1 <!DOCTYPE HTML PUBLIC "-//W3C//DTD XHTML 1.1//EN" "http://www.w3.org/TR/xhtml11/DTD/xhtml11.dtd">
4 <title>Stacker: An Example Of Using LLVM</title>
5 <link rel="stylesheet" href="llvm.css" type="text/css">
8 <div class="doc_title">Stacker: An Example Of Using LLVM</div>
11 <li><a href="#abstract">Abstract</a></li>
12 <li><a href="#introduction">Introduction</a></li>
13 <li><a href="#lessons">Lessons I Learned About LLVM</a>
15 <li><a href="#value">Everything's a Value!</a></li>
16 <li><a href="#terminate">Terminate Those Blocks!</a></li>
17 <li><a href="#blocks">Concrete Blocks</a></li>
18 <li><a href="#push_back">push_back Is Your Friend</a></li>
19 <li><a href="#gep">The Wily GetElementPtrInst</a></li>
20 <li><a href="#linkage">Getting Linkage Types Right</a></li>
21 <li><a href="#constants">Constants Are Easier Than That!</a></li>
24 <li><a href="#lexicon">The Stacker Lexicon</a>
26 <li><a href="#stack">The Stack</a>
27 <li><a href="#punctuation">Punctuation</a>
28 <li><a href="#comments">Comments</a>
29 <li><a href="#literals">Literals</a>
30 <li><a href="#words">Words</a>
31 <li><a href="style">Standard Style</a>
32 <li><a href="#builtins">Built-Ins</a>
35 <li><a href="#example">Prime: A Complete Example</a></li>
36 <li><a href="#internal">Internal Code Details</a>
38 <li><a href="#directory">The Directory Structure </a></li>
39 <li><a href="#lexer">The Lexer</a></li>
40 <li><a href="#parser">The Parser</a></li>
41 <li><a href="#compiler">The Compiler</a></li>
42 <li><a href="#runtime">The Runtime</a></li>
43 <li><a href="#driver">Compiler Driver</a></li>
44 <li><a href="#tests">Test Programs</a></li>
45 <li><a href="#exercise">Exercise</a></li>
46 <li><a href="#todo">Things Remaining To Be Done</a></li>
50 <div class="doc_text">
51 <p><b>Written by <a href="mailto:rspencer@x10sys.com">Reid Spencer</a> </b></p>
55 <!-- ======================================================================= -->
56 <div class="doc_section"> <a name="abstract">Abstract </a></div>
57 <div class="doc_text">
58 <p>This document is another way to learn about LLVM. Unlike the
59 <a href="LangRef.html">LLVM Reference Manual</a> or
60 <a href="ProgrammersManual.html">LLVM Programmer's Manual</a>, here we learn
61 about LLVM through the experience of creating a simple programming language
62 named Stacker. Stacker was invented specifically as a demonstration of
63 LLVM. The emphasis in this document is not on describing the
64 intricacies of LLVM itself, but on how to use it to build your own
67 <!-- ======================================================================= -->
68 <div class="doc_section"> <a name="introduction">Introduction</a> </div>
69 <div class="doc_text">
70 <p>Amongst other things, LLVM is a platform for compiler writers.
71 Because of its exceptionally clean and small IR (intermediate
72 representation), compiler writing with LLVM is much easier than with
73 other system. As proof, I wrote the entire compiler (language definition,
74 lexer, parser, code generator, etc.) in about <em>four days</em>!
75 That's important to know because it shows how quickly you can get a new
76 language running when using LLVM. Furthermore, this was the <em >first</em>
77 language the author ever created using LLVM. The learning curve is
78 included in that four days.</p>
79 <p>The language described here, Stacker, is Forth-like. Programs
80 are simple collections of word definitions and the only thing definitions
81 can do is manipulate a stack or generate I/O. Stacker is not a "real"
82 programming language; its very simple. Although it is computationally
83 complete, you wouldn't use it for your next big project. However,
84 the fact that it is complete, its simple, and it <em>doesn't</em> have
85 a C-like syntax make it useful for demonstration purposes. It shows
86 that LLVM could be applied to a wide variety of languages.</p>
87 <p>The basic notions behind stacker is very simple. There's a stack of
88 integers (or character pointers) that the program manipulates. Pretty
89 much the only thing the program can do is manipulate the stack and do
90 some limited I/O operations. The language provides you with several
91 built-in words that manipulate the stack in interesting ways. To get
92 your feet wet, here's how you write the traditional "Hello, World"
93 program in Stacker:</p>
94 <p><code>: hello_world "Hello, World!" >s DROP CR ;<br>
95 : MAIN hello_world ;<br></code></p>
96 <p>This has two "definitions" (Stacker manipulates words, not
97 functions and words have definitions): <code>MAIN</code> and <code>
98 hello_world</code>. The <code>MAIN</code> definition is standard, it
99 tells Stacker where to start. Here, <code>MAIN</code> is defined to
100 simply invoke the word <code>hello_world</code>. The
101 <code>hello_world</code> definition tells stacker to push the
102 <code>"Hello, World!"</code> string onto the stack, print it out
103 (<code>>s</code>), pop it off the stack (<code>DROP</code>), and
104 finally print a carriage return (<code>CR</code>). Although
105 <code>hello_world</code> uses the stack, its net effect is null. Well
106 written Stacker definitions have that characteristic. </p>
107 <p>Exercise for the reader: how could you make this a one line program?</p>
109 <!-- ======================================================================= -->
110 <div class="doc_section"><a name="lessons"></a>Lessons I Learned About LLVM</div>
111 <div class="doc_text">
112 <p>Stacker was written for two purposes: </p>
114 <li>to get the author over the learning curve, and</li>
115 <li>to provide a simple example of how to write a compiler using LLVM.</li>
117 <p>During the development of Stacker, many lessons about LLVM were
118 learned. Those lessons are described in the following subsections.<p>
120 <!-- ======================================================================= -->
121 <div class="doc_subsection"><a name="value"></a>Everything's a Value!</div>
122 <div class="doc_text">
123 <p>Although I knew that LLVM uses a Single Static Assignment (SSA) format,
124 it wasn't obvious to me how prevalent this idea was in LLVM until I really
125 started using it. Reading the <a href="ProgrammersManual.html">
126 Programmer's Manual</a> and <a href="LangRef.html">Language Reference</a>
127 I noted that most of the important LLVM IR (Intermediate Representation) C++
128 classes were derived from the Value class. The full power of that simple
129 design only became fully understood once I started constructing executable
130 expressions for Stacker.</p>
131 <p>This really makes your programming go faster. Think about compiling code
132 for the following C/C++ expression: <code>(a|b)*((x+1)/(y+1))</code>. Assuming
133 the values are on the stack in the order a, b, x, y, this could be
134 expressed in stacker as: <code>1 + SWAP 1 + / ROT2 OR *</code>.
135 You could write a function using LLVM that computes this expression like this: </p>
138 expression(BasicBlock* bb, Value* a, Value* b, Value* x, Value* y )
140 Instruction* tail = bb->getTerminator();
141 ConstantSInt* one = ConstantSInt::get( Type::IntTy, 1);
142 BinaryOperator* or1 =
143 BinaryOperator::create( Instruction::Or, a, b, "", tail );
144 BinaryOperator* add1 =
145 BinaryOperator::create( Instruction::Add, x, one, "", tail );
146 BinaryOperator* add2 =
147 BinaryOperator::create( Instruction::Add, y, one, "", tail );
148 BinaryOperator* div1 =
149 BinaryOperator::create( Instruction::Div, add1, add2, "", tail);
150 BinaryOperator* mult1 =
151 BinaryOperator::create( Instruction::Mul, or1, div1, "", tail );
156 <p>"Okay, big deal," you say? It is a big deal. Here's why. Note that I didn't
157 have to tell this function which kinds of Values are being passed in. They could be
158 <code>Instruction</code>s, <code>Constant</code>s, <code>GlobalVariable</code>s, or
159 any of the other subclasses of <code>Value</code> that LLVM supports.
160 Furthermore, if you specify Values that are incorrect for this sequence of
161 operations, LLVM will either notice right away (at compilation time) or the LLVM
162 Verifier will pick up the inconsistency when the compiler runs. In either case
163 LLVM prevents you from making a type error that gets passed through to the
164 generated program. This <em>really</em> helps you write a compiler that
165 always generates correct code!<p>
166 <p>The second point is that we don't have to worry about branching, registers,
167 stack variables, saving partial results, etc. The instructions we create
168 <em>are</em> the values we use. Note that all that was created in the above
169 code is a Constant value and five operators. Each of the instructions <em>is</em>
170 the resulting value of that instruction. This saves a lot of time.</p>
171 <p>The lesson is this: <em>SSA form is very powerful: there is no difference
172 between a value and the instruction that created it.</em> This is fully
173 enforced by the LLVM IR. Use it to your best advantage.</p>
175 <!-- ======================================================================= -->
176 <div class="doc_subsection"><a name="terminate"></a>Terminate Those Blocks!</div>
177 <div class="doc_text">
178 <p>I had to learn about terminating blocks the hard way: using the debugger
179 to figure out what the LLVM verifier was trying to tell me and begging for
180 help on the LLVMdev mailing list. I hope you avoid this experience.</p>
181 <p>Emblazon this rule in your mind:</p>
183 <li><em>All</em> <code>BasicBlock</code>s in your compiler <b>must</b> be
184 terminated with a terminating instruction (branch, return, etc.).
187 <p>Terminating instructions are a semantic requirement of the LLVM IR. There
188 is no facility for implicitly chaining together blocks placed into a function
189 in the order they occur. Indeed, in the general case, blocks will not be
190 added to the function in the order of execution because of the recursive
191 way compilers are written.</p>
192 <p>Furthermore, if you don't terminate your blocks, your compiler code will
193 compile just fine. You won't find out about the problem until you're running
194 the compiler and the module you just created fails on the LLVM Verifier.</p>
196 <!-- ======================================================================= -->
197 <div class="doc_subsection"><a name="blocks"></a>Concrete Blocks</div>
198 <div class="doc_text">
199 <p>After a little initial fumbling around, I quickly caught on to how blocks
200 should be constructed. In general, here's what I learned:
202 <li><em>Create your blocks early.</em> While writing your compiler, you
203 will encounter several situations where you know apriori that you will
204 need several blocks. For example, if-then-else, switch, while and for
205 statements in C/C++ all need multiple blocks for expression in LVVM.
206 The rule is, create them early.</li>
207 <li><em>Terminate your blocks early.</em> This just reduces the chances
208 that you forget to terminate your blocks which is required (go
209 <a href="#terminate">here</a> for more).
210 <li><em>Use getTerminator() for instruction insertion.</em> I noticed early on
211 that many of the constructors for the Instruction classes take an optional
212 <code>insert_before</code> argument. At first, I thought this was a mistake
213 because clearly the normal mode of inserting instructions would be one at
214 a time <em>after</em> some other instruction, not <em>before</em>. However,
215 if you hold on to your terminating instruction (or use the handy dandy
216 <code>getTerminator()</code> method on a <code>BasicBlock</code>), it can
217 always be used as the <code>insert_before</code> argument to your instruction
218 constructors. This causes the instruction to automatically be inserted in
219 the RightPlace™ place, just before the terminating instruction. The
220 nice thing about this design is that you can pass blocks around and insert
221 new instructions into them without ever knowing what instructions came
222 before. This makes for some very clean compiler design.</li>
224 <p>The foregoing is such an important principal, its worth making an idiom:</p>
226 BasicBlock* bb = new BasicBlock();</li>
227 bb->getInstList().push_back( new Branch( ... ) );
228 new Instruction(..., bb->getTerminator() );
230 <p>To make this clear, consider the typical if-then-else statement
231 (see StackerCompiler::handle_if() method). We can set this up
232 in a single function using LLVM in the following way: </p>
234 using namespace llvm;
236 MyCompiler::handle_if( BasicBlock* bb, SetCondInst* condition )
238 // Create the blocks to contain code in the structure of if/then/else
239 BasicBlock* then_bb = new BasicBlock();
240 BasicBlock* else_bb = new BasicBlock();
241 BasicBlock* exit_bb = new BasicBlock();
243 // Insert the branch instruction for the "if"
244 bb->getInstList().push_back( new BranchInst( then_bb, else_bb, condition ) );
246 // Set up the terminating instructions
247 then->getInstList().push_back( new BranchInst( exit_bb ) );
248 else->getInstList().push_back( new BranchInst( exit_bb ) );
250 // Fill in the then part .. details excised for brevity
251 this->fill_in( then_bb );
253 // Fill in the else part .. details excised for brevity
254 this->fill_in( else_bb );
256 // Return a block to the caller that can be filled in with the code
257 // that follows the if/then/else construct.
261 <p>Presumably in the foregoing, the calls to the "fill_in" method would add
262 the instructions for the "then" and "else" parts. They would use the third part
263 of the idiom almost exclusively (inserting new instructions before the
264 terminator). Furthermore, they could even recurse back to <code>handle_if</code>
265 should they encounter another if/then/else statement and it will just work.</p>
266 <p>Note how cleanly this all works out. In particular, the push_back methods on
267 the <code>BasicBlock</code>'s instruction list. These are lists of type
268 <code>Instruction</code> (which is also of type <code>Value</code>). To create
269 the "if" branch we merely instantiate a <code>BranchInst</code> that takes as
270 arguments the blocks to branch to and the condition to branch on. The
271 <code>BasicBlock</code> objects act like branch labels! This new
272 <code>BranchInst</code> terminates the <code>BasicBlock</code> provided
273 as an argument. To give the caller a way to keep inserting after calling
274 <code>handle_if</code> we create an <code>exit_bb</code> block which is returned
275 to the caller. Note that the <code>exit_bb</code> block is used as the
276 terminator for both the <code>then_bb</code> and the <code>else_bb</code>
277 blocks. This guarantees that no matter what else <code>handle_if</code>
278 or <code>fill_in</code> does, they end up at the <code>exit_bb</code> block.
281 <!-- ======================================================================= -->
282 <div class="doc_subsection"><a name="push_back"></a>push_back Is Your Friend</div>
283 <div class="doc_text">
285 One of the first things I noticed is the frequent use of the "push_back"
286 method on the various lists. This is so common that it is worth mentioning.
287 The "push_back" inserts a value into an STL list, vector, array, etc. at the
288 end. The method might have also been named "insert_tail" or "append".
289 Althought I've used STL quite frequently, my use of push_back wasn't very
290 high in other programs. In LLVM, you'll use it all the time.
293 <!-- ======================================================================= -->
294 <div class="doc_subsection"><a name="gep"></a>The Wily GetElementPtrInst</div>
295 <div class="doc_text">
297 It took a little getting used to and several rounds of postings to the LLVM
298 mail list to wrap my head around this instruction correctly. Even though I had
299 read the Language Reference and Programmer's Manual a couple times each, I still
300 missed a few <em>very</em> key points:
303 <li>GetElementPtrInst gives you back a Value for the last thing indexed</em>
304 <li>All global variables in LLVM are <em>pointers</em>.
305 <li>Pointers must also be dereferenced with the GetElementPtrInst instruction.
307 <p>This means that when you look up an element in the global variable (assuming
308 its a struct or array), you <em>must</em> deference the pointer first! For many
309 things, this leads to the idiom:
312 std::vector<Value*> index_vector;
313 index_vector.push_back( ConstantSInt::get( Type::LongTy, 0 );
314 // ... push other indices ...
315 GetElementPtrInst* gep = new GetElementPtrInst( ptr, index_vector );
317 <p>For example, suppose we have a global variable whose type is [24 x int]. The
318 variable itself represents a <em>pointer</em> to that array. To subscript the
319 array, we need two indices, not just one. The first index (0) dereferences the
320 pointer. The second index subscripts the array. If you're a "C" programmer, this
321 will run against your grain because you'll naturally think of the global array
322 variable and the address of its first element as the same. That tripped me up
323 for a while until I realized that they really do differ .. by <em>type</em>.
324 Remember that LLVM is strongly typed. Everything has a type.
325 The "type" of the global variable is [24 x int]*. That is, its
326 a pointer to an array of 24 ints. When you dereference that global variable with
327 a single (0) index, you now have a "[24 x int]" type. Although
328 the pointer value of the dereferenced global and the address of the zero'th element
329 in the array will be the same, they differ in their type. The zero'th element has
330 type "int" while the pointer value has type "[24 x int]".</p>
331 <p>Get this one aspect of LLVM right in your head and you'll save yourself
332 a lot of compiler writing headaches down the road.</p>
334 <!-- ======================================================================= -->
335 <div class="doc_subsection"><a name="linkage"></a>Getting Linkage Types Right</div>
336 <div class="doc_text">
337 <p>Linkage types in LLVM can be a little confusing, especially if your compiler
338 writing mind has affixed firm concepts to particular words like "weak",
339 "external", "global", "linkonce", etc. LLVM does <em>not</em> use the precise
340 definitions of say ELF or GCC even though they share common terms. To be fair,
341 the concepts are related and similar but not precisely the same. This can lead
342 you to think you know what a linkage type represents but in fact it is slightly
343 different. I recommend you read the
344 <a href="LangRef.html#linkage"> Language Reference on this topic</a> very
345 carefully. Then, read it again.<p>
346 <p>Here are some handy tips that I discovered along the way:</p>
348 <li><em>Unitialized means external.</em> That is, the symbol is declared in the current
349 module and can be used by that module but it is not defined by that module.</li>
350 <li><em>Setting an initializer changes a global' linkage type.</em> Setting an
351 initializer changes a global's linkage type from whatever it was to a normal,
352 defind global (not external). You'll need to call the setLinkage() method to
353 reset it if you specify the initializer after the GlobalValue has been constructed.
354 This is important for LinkOnce and Weak linkage types.</li>
355 <li><em>Appending linkage can keep track of things.</em> Appending linkage can
356 be used to keep track of compilation information at runtime. It could be used,
357 for example, to build a full table of all the C++ virtual tables or hold the
358 C++ RTTI data, or whatever. Appending linkage can only be applied to arrays.
359 All arrays with the same name in each module are concatenated together at link
363 <!-- ======================================================================= -->
364 <div class="doc_subsection"><a name="constants"></a>Constants Are Easier Than That!</div>
365 <div class="doc_text">
367 Constants in LLVM took a little getting used to until I discovered a few utility
368 functions in the LLVM IR that make things easier. Here's what I learned: </p>
370 <li>Constants are Values like anything else and can be operands of instructions</li>
371 <li>Integer constants, frequently needed can be created using the static "get"
372 methods of the ConstantInt, ConstantSInt, and ConstantUInt classes. The nice thing
373 about these is that you can "get" any kind of integer quickly.</li>
374 <li>There's a special method on Constant class which allows you to get the null
375 constant for <em>any</em> type. This is really handy for initializing large
376 arrays or structures, etc.</li>
379 <!-- ======================================================================= -->
380 <div class="doc_section"> <a name="lexicon">The Stacker Lexicon</a></div>
381 <div class="doc_text"><p>This section describes the Stacker language</p></div>
382 <div class="doc_subsection"><a name="stack"></a>The Stack</div>
383 <div class="doc_text">
384 <p>Stacker definitions define what they do to the global stack. Before
385 proceeding, a few words about the stack are in order. The stack is simply
386 a global array of 32-bit integers or pointers. A global index keeps track
387 of the location of the top of the stack. All of this is hidden from the
388 programmer but it needs to be noted because it is the foundation of the
389 conceptual programming model for Stacker. When you write a definition,
390 you are, essentially, saying how you want that definition to manipulate
391 the global stack.</p>
392 <p>Manipulating the stack can be quite hazardous. There is no distinction
393 given and no checking for the various types of values that can be placed
394 on the stack. Automatic coercion between types is performed. In many
395 cases this is useful. For example, a boolean value placed on the stack
396 can be interpreted as an integer with good results. However, using a
397 word that interprets that boolean value as a pointer to a string to
398 print out will almost always yield a crash. Stacker simply leaves it
399 to the programmer to get it right without any interference or hindering
400 on interpretation of the stack values. You've been warned. :) </p>
402 <!-- ======================================================================= -->
403 <div class="doc_subsection"> <a name="punctuation"></a>Punctuation</div>
404 <div class="doc_text">
405 <p>Punctuation in Stacker is very simple. The colon and semi-colon
406 characters are used to introduce and terminate a definition
407 (respectively). Except for <em>FORWARD</em> declarations, definitions
408 are all you can specify in Stacker. Definitions are read left to right.
409 Immediately after the colon comes the name of the word being defined.
410 The remaining words in the definition specify what the word does. The definition
411 is terminated by a semi-colon.</p>
412 <p>So, your typical definition will have the form:</p>
413 <pre><code>: name ... ;</code></pre>
414 <p>The <code>name</code> is up to you but it must start with a letter and contain
415 only letters numbers and underscore. Names are case sensitive and must not be
416 the same as the name of a built-in word. The <code>...</code> is replaced by
417 the stack manipulting words that you wish define <code>name</code> as. <p>
419 <!-- ======================================================================= -->
420 <div class="doc_subsection"><a name="comments"></a>Comments</div>
421 <div class="doc_text">
422 <p>Stacker supports two types of comments. A hash mark (#) starts a comment
423 that extends to the end of the line. It is identical to the kind of comments
424 commonly used in shell scripts. A pair of parentheses also surround a comment.
425 In both cases, the content of the comment is ignored by the Stacker compiler. The
426 following does nothing in Stacker.
429 # This is a comment to end of line
430 ( This is an enclosed comment )
432 <p>See the <a href="#example">example</a> program to see comments in use in
435 <!-- ======================================================================= -->
436 <div class="doc_subsection"><a name="literals"></a>Literals</div>
437 <div class="doc_text">
438 <p>There are three kinds of literal values in Stacker. Integer, Strings,
439 and Booleans. In each case, the stack operation is to simply push the
440 value onto the stack. So, for example:<br/>
441 <code> 42 " is the answer." TRUE </code><br/>
442 will push three values onto the stack: the integer 42, the
443 string " is the answer." and the boolean TRUE.</p>
445 <!-- ======================================================================= -->
446 <div class="doc_subsection"><a name="words"></a>Words</div>
447 <div class="doc_text">
448 <p>Each definition in Stacker is composed of a set of words. Words are
449 read and executed in order from left to right. There is very little
450 checking in Stacker to make sure you're doing the right thing with
451 the stack. It is assumed that the programmer knows how the stack
452 transformation he applies will affect the program.</p>
453 <p>Words in a definition come in two flavors: built-in and programmer
454 defined. Simply mentioning the name of a previously defined or declared
455 programmer-defined word causes that word's stack actions to be invoked. It
456 is somewhat like a function call in other languages. The built-in
457 words have various effects, described <a href="#builtins">below</a>.</p>
458 <p>Sometimes you need to call a word before it is defined. For this, you can
459 use the <code>FORWARD</code> declaration. It looks like this:</p>
460 <p><code>FORWARD name ;</code></p>
461 <p>This simply states to Stacker that "name" is the name of a definition
462 that is defined elsewhere. Generally it means the definition can be found
463 "forward" in the file. But, it doesn't have to be in the current compilation
464 unit. Anything declared with <code>FORWARD</code> is an external symbol for
467 <!-- ======================================================================= -->
468 <div class="doc_subsection"><a name="builtins"></a>Built In Words</div>
469 <div class="doc_text">
470 <p>The built-in words of the Stacker language are put in several groups
471 depending on what they do. The groups are as follows:</p>
473 <li><em>Logical</em>These words provide the logical operations for
474 comparing stack operands.<br/>The words are: < > <= >=
475 = <> true false.</li>
476 <li><em>Bitwise</em>These words perform bitwise computations on
477 their operands. <br/> The words are: << >> XOR AND NOT</li>
478 <li><em>Arithmetic</em>These words perform arithmetic computations on
479 their operands. <br/> The words are: ABS NEG + - * / MOD */ ++ -- MIN MAX</li>
480 <li><em>Stack</em>These words manipulate the stack directly by moving
481 its elements around.<br/> The words are: DROP DROP2 NIP NIP2 DUP DUP2
482 SWAP SWAP2 OVER OVER2 ROT ROT2 RROT RROT2 TUCK TUCK2 PICK SELECT ROLL</li>
483 <li><em>Memory</em>These words allocate, free and manipulate memory
484 areas outside the stack.<br/>The words are: MALLOC FREE GET PUT</li>
485 <li><em>Control</em>These words alter the normal left to right flow
486 of execution.<br/>The words are: IF ELSE ENDIF WHILE END RETURN EXIT RECURSE</li>
487 <li><em>I/O</em> These words perform output on the standard output
488 and input on the standard input. No other I/O is possible in Stacker.
489 <br/>The words are: SPACE TAB CR >s >d >c <s <d <c.</li>
491 <p>While you may be familiar with many of these operations from other
492 programming languages, a careful review of their semantics is important
493 for correct programming in Stacker. Of most importance is the effect
494 that each of these built-in words has on the global stack. The effect is
495 not always intuitive. To better describe the effects, we'll borrow from Forth the idiom of
496 describing the effect on the stack with:</p>
497 <p><code> BEFORE -- AFTER </code></p>
498 <p>That is, to the left of the -- is a representation of the stack before
499 the operation. To the right of the -- is a representation of the stack
500 after the operation. In the table below that describes the operation of
501 each of the built in words, we will denote the elements of the stack
502 using the following construction:</p>
504 <li><em>b</em> - a boolean truth value</li>
505 <li><em>w</em> - a normal integer valued word.</li>
506 <li><em>s</em> - a pointer to a string value</li>
507 <li><em>p</em> - a pointer to a malloc'd memory block</li>
510 <div class="doc_text" >
511 <table class="doc_table" style="border: 2px solid blue; border-collapse: collapse;" >
512 <tr class="doc_table"><td colspan="4" style="border: 2px solid blue">Definition Of Operation Of Built In Words</td></tr>
513 <tr class="doc_table"><td colspan="4" style="border: 2px solid blue"><b>LOGICAL OPERATIONS</b></td></tr>
514 <tr class="doc_table">
515 <td style="border: 2px solid blue"><u>Word</u></td>
516 <td style="border: 2px solid blue"><u>Name</u></td>
517 <td style="border: 2px solid blue"><u>Operation</u></td>
518 <td style="border: 2px solid blue"><u>Description</u></td>
520 <tr class="doc_table"><td style="border: 2px solid blue"><</td>
521 <td style="border: 2px solid blue">LT</td>
522 <td style="border: 2px solid blue">w1 w2 -- b</td>
523 <td style="border: 2px solid blue">Two values (w1 and w2) are popped off the stack and
524 compared. If w1 is less than w2, TRUE is pushed back on
525 the stack, otherwise FALSE is pushed back on the stack.</td>
527 <tr><td style="border: 2px solid blue">></td>
528 <td style="border: 2px solid blue">GT</td>
529 <td style="border: 2px solid blue">w1 w2 -- b</td>
530 <td style="border: 2px solid blue">Two values (w1 and w2) are popped off the stack and
531 compared. If w1 is greater than w2, TRUE is pushed back on
532 the stack, otherwise FALSE is pushed back on the stack.</td>
534 <tr><td style="border: 2px solid blue">>=</td>
535 <td style="border: 2px solid blue">GE</td>
536 <td style="border: 2px solid blue">w1 w2 -- b</td>
537 <td style="border: 2px solid blue">Two values (w1 and w2) are popped off the stack and
538 compared. If w1 is greater than or equal to w2, TRUE is
539 pushed back on the stack, otherwise FALSE is pushed back
542 <tr><td style="border: 2px solid blue"><=</td>
543 <td style="border: 2px solid blue">LE</td>
544 <td style="border: 2px solid blue">w1 w2 -- b</td>
545 <td style="border: 2px solid blue">Two values (w1 and w2) are popped off the stack and
546 compared. If w1 is less than or equal to w2, TRUE is
547 pushed back on the stack, otherwise FALSE is pushed back
550 <tr><td style="border: 2px solid blue">=</td>
551 <td style="border: 2px solid blue">EQ</td>
552 <td style="border: 2px solid blue">w1 w2 -- b</td>
553 <td style="border: 2px solid blue">Two values (w1 and w2) are popped off the stack and
554 compared. If w1 is equal to w2, TRUE is
555 pushed back on the stack, otherwise FALSE is pushed back
558 <tr><td style="border: 2px solid blue"><></td>
559 <td style="border: 2px solid blue">NE</td>
560 <td style="border: 2px solid blue">w1 w2 -- b</td>
561 <td style="border: 2px solid blue">Two values (w1 and w2) are popped off the stack and
562 compared. If w1 is equal to w2, TRUE is
563 pushed back on the stack, otherwise FALSE is pushed back
566 <tr><td style="border: 2px solid blue">FALSE</td>
567 <td style="border: 2px solid blue">FALSE</td>
568 <td style="border: 2px solid blue"> -- b</td>
569 <td style="border: 2px solid blue">The boolean value FALSE (0) is pushed onto the stack.</td>
571 <tr><td style="border: 2px solid blue">TRUE</td>
572 <td style="border: 2px solid blue">TRUE</td>
573 <td style="border: 2px solid blue"> -- b</td>
574 <td style="border: 2px solid blue">The boolean value TRUE (-1) is pushed onto the stack.</td>
576 <tr><td colspan="4"><b>BITWISE OPERATORS</b></td></tr>
578 <td style="border: 2px solid blue"><u>Word</u></td>
579 <td style="border: 2px solid blue"><u>Name</u></td>
580 <td style="border: 2px solid blue"><u>Operation</u></td>
581 <td style="border: 2px solid blue"><u>Description</u></td>
583 <tr><td style="border: 2px solid blue"><<</td>
584 <td style="border: 2px solid blue">SHL</td>
585 <td style="border: 2px solid blue">w1 w2 -- w1<<w2</td>
586 <td style="border: 2px solid blue">Two values (w1 and w2) are popped off the stack. The w2
587 operand is shifted left by the number of bits given by the
588 w1 operand. The result is pushed back to the stack.</td>
590 <tr><td style="border: 2px solid blue">>></td>
591 <td style="border: 2px solid blue">SHR</td>
592 <td style="border: 2px solid blue">w1 w2 -- w1>>w2</td>
593 <td style="border: 2px solid blue">Two values (w1 and w2) are popped off the stack. The w2
594 operand is shifted right by the number of bits given by the
595 w1 operand. The result is pushed back to the stack.</td>
597 <tr><td style="border: 2px solid blue">OR</td>
598 <td style="border: 2px solid blue">OR</td>
599 <td style="border: 2px solid blue">w1 w2 -- w2|w1</td>
600 <td style="border: 2px solid blue">Two values (w1 and w2) are popped off the stack. The values
601 are bitwise OR'd together and pushed back on the stack. This is
602 not a logical OR. The sequence 1 2 OR yields 3 not 1.</td>
604 <tr><td style="border: 2px solid blue">AND</td>
605 <td style="border: 2px solid blue">AND</td>
606 <td style="border: 2px solid blue">w1 w2 -- w2&w1</td>
607 <td style="border: 2px solid blue">Two values (w1 and w2) are popped off the stack. The values
608 are bitwise AND'd together and pushed back on the stack. This is
609 not a logical AND. The sequence 1 2 AND yields 0 not 1.</td>
611 <tr><td style="border: 2px solid blue">XOR</td>
612 <td style="border: 2px solid blue">XOR</td>
613 <td style="border: 2px solid blue">w1 w2 -- w2^w1</td>
614 <td style="border: 2px solid blue">Two values (w1 and w2) are popped off the stack. The values
615 are bitwise exclusive OR'd together and pushed back on the stack.
616 For example, The sequence 1 3 XOR yields 2.</td>
618 <tr><td colspan="4"><b>ARITHMETIC OPERATORS</b></td></tr>
620 <td style="border: 2px solid blue"><u>Word</u></td>
621 <td style="border: 2px solid blue"><u>Name</u></td>
622 <td style="border: 2px solid blue"><u>Operation</u></td>
623 <td style="border: 2px solid blue"><u>Description</u></td>
625 <tr><td style="border: 2px solid blue">ABS</td>
626 <td style="border: 2px solid blue">ABS</td>
627 <td style="border: 2px solid blue">w -- |w|</td>
628 <td style="border: 2px solid blue">One value s popped off the stack; its absolute value is computed
629 and then pushed onto the stack. If w1 is -1 then w2 is 1. If w1 is
630 1 then w2 is also 1.</td>
632 <tr><td style="border: 2px solid blue">NEG</td>
633 <td style="border: 2px solid blue">NEG</td>
634 <td style="border: 2px solid blue">w -- -w</td>
635 <td style="border: 2px solid blue">One value is popped off the stack which is negated and then
636 pushed back onto the stack. If w1 is -1 then w2 is 1. If w1 is
637 1 then w2 is -1.</td>
639 <tr><td style="border: 2px solid blue"> + </td>
640 <td style="border: 2px solid blue">ADD</td>
641 <td style="border: 2px solid blue">w1 w2 -- w2+w1</td>
642 <td style="border: 2px solid blue">Two values are popped off the stack. Their sum is pushed back
645 <tr><td style="border: 2px solid blue"> - </td>
646 <td style="border: 2px solid blue">SUB</td>
647 <td style="border: 2px solid blue">w1 w2 -- w2-w1</td>
648 <td style="border: 2px solid blue">Two values are popped off the stack. Their difference is pushed back
651 <tr><td style="border: 2px solid blue"> * </td>
652 <td style="border: 2px solid blue">MUL</td>
653 <td style="border: 2px solid blue">w1 w2 -- w2*w1</td>
654 <td style="border: 2px solid blue">Two values are popped off the stack. Their product is pushed back
657 <tr><td style="border: 2px solid blue"> / </td>
658 <td style="border: 2px solid blue">DIV</td>
659 <td style="border: 2px solid blue">w1 w2 -- w2/w1</td>
660 <td style="border: 2px solid blue">Two values are popped off the stack. Their quotient is pushed back
663 <tr><td style="border: 2px solid blue">MOD</td>
664 <td style="border: 2px solid blue">MOD</td>
665 <td style="border: 2px solid blue">w1 w2 -- w2%w1</td>
666 <td style="border: 2px solid blue">Two values are popped off the stack. Their remainder after division
667 of w1 by w2 is pushed back onto the stack</td>
669 <tr><td style="border: 2px solid blue"> */ </td>
670 <td style="border: 2px solid blue">STAR_SLAH</td>
671 <td style="border: 2px solid blue">w1 w2 w3 -- (w3*w2)/w1</td>
672 <td style="border: 2px solid blue">Three values are popped off the stack. The product of w1 and w2 is
673 divided by w3. The result is pushed back onto the stack.</td>
675 <tr><td style="border: 2px solid blue"> ++ </td>
676 <td style="border: 2px solid blue">INCR</td>
677 <td style="border: 2px solid blue">w -- w+1</td>
678 <td style="border: 2px solid blue">One value is popped off the stack. It is incremented by one and then
679 pushed back onto the stack.</td>
681 <tr><td style="border: 2px solid blue"> -- </td>
682 <td style="border: 2px solid blue">DECR</td>
683 <td style="border: 2px solid blue">w -- w-1</td>
684 <td style="border: 2px solid blue">One value is popped off the stack. It is decremented by one and then
685 pushed back onto the stack.</td>
687 <tr><td style="border: 2px solid blue">MIN</td>
688 <td style="border: 2px solid blue">MIN</td>
689 <td style="border: 2px solid blue">w1 w2 -- (w2<w1?w2:w1)</td>
690 <td style="border: 2px solid blue">Two values are popped off the stack. The larger one is pushed back
693 <tr><td style="border: 2px solid blue">MAX</td>
694 <td style="border: 2px solid blue">MAX</td>
695 <td style="border: 2px solid blue">w1 w2 -- (w2>w1?w2:w1)</td>
696 <td style="border: 2px solid blue">Two values are popped off the stack. The larger value is pushed back
699 <tr><td colspan="4"><b>STACK MANIPULATION OPERATORS</b></td></tr>
701 <td style="border: 2px solid blue"><u>Word</u></td>
702 <td style="border: 2px solid blue"><u>Name</u></td>
703 <td style="border: 2px solid blue"><u>Operation</u></td>
704 <td style="border: 2px solid blue"><u>Description</u></td>
706 <tr><td style="border: 2px solid blue">DROP</td>
707 <td style="border: 2px solid blue">DROP</td>
708 <td style="border: 2px solid blue">w -- </td>
709 <td style="border: 2px solid blue">One value is popped off the stack.</td>
711 <tr><td style="border: 2px solid blue">DROP2</td>
712 <td style="border: 2px solid blue">DROP2</td>
713 <td style="border: 2px solid blue">w1 w2 -- </td>
714 <td style="border: 2px solid blue">Two values are popped off the stack.</td>
716 <tr><td style="border: 2px solid blue">NIP</td>
717 <td style="border: 2px solid blue">NIP</td>
718 <td style="border: 2px solid blue">w1 w2 -- w2</td>
719 <td style="border: 2px solid blue">The second value on the stack is removed from the stack. That is,
720 a value is popped off the stack and retained. Then a second value is
721 popped and the retained value is pushed.</td>
723 <tr><td style="border: 2px solid blue">NIP2</td>
724 <td style="border: 2px solid blue">NIP2</td>
725 <td style="border: 2px solid blue">w1 w2 w3 w4 -- w3 w4</td>
726 <td style="border: 2px solid blue">The third and fourth values on the stack are removed from it. That is,
727 two values are popped and retained. Then two more values are popped and
728 the two retained values are pushed back on.</td>
730 <tr><td style="border: 2px solid blue">DUP</td>
731 <td style="border: 2px solid blue">DUP</td>
732 <td style="border: 2px solid blue">w1 -- w1 w1</td>
733 <td style="border: 2px solid blue">One value is popped off the stack. That value is then pushed onto
734 the stack twice to duplicate the top stack vaue.</td>
736 <tr><td style="border: 2px solid blue">DUP2</td>
737 <td style="border: 2px solid blue">DUP2</td>
738 <td style="border: 2px solid blue">w1 w2 -- w1 w2 w1 w2</td>
739 <td style="border: 2px solid blue">The top two values on the stack are duplicated. That is, two vaues
740 are popped off the stack. They are alternately pushed back on the
741 stack twice each.</td>
743 <tr><td style="border: 2px solid blue">SWAP</td>
744 <td style="border: 2px solid blue">SWAP</td>
745 <td style="border: 2px solid blue">w1 w2 -- w2 w1</td>
746 <td style="border: 2px solid blue">The top two stack items are reversed in their order. That is, two
747 values are popped off the stack and pushed back onto the stack in
748 the opposite order they were popped.</td>
750 <tr><td style="border: 2px solid blue">SWAP2</td>
751 <td style="border: 2px solid blue">SWAP2</td>
752 <td style="border: 2px solid blue">w1 w2 w3 w4 -- w3 w4 w2 w1</td>
753 <td style="border: 2px solid blue">The top four stack items are swapped in pairs. That is, two values
754 are popped and retained. Then, two more values are popped and retained.
755 The values are pushed back onto the stack in the reverse order but
758 <tr><td style="border: 2px solid blue">OVER</td>
759 <td style="border: 2px solid blue">OVER</td>
760 <td style="border: 2px solid blue">w1 w2-- w1 w2 w1</td>
761 <td style="border: 2px solid blue">Two values are popped from the stack. They are pushed back
762 onto the stack in the order w1 w2 w1. This seems to cause the
763 top stack element to be duplicated "over" the next value.</td>
765 <tr><td style="border: 2px solid blue">OVER2</td>
766 <td style="border: 2px solid blue">OVER2</td>
767 <td style="border: 2px solid blue">w1 w2 w3 w4 -- w1 w2 w3 w4 w1 w2</td>
768 <td style="border: 2px solid blue">The third and fourth values on the stack are replicated onto the
769 top of the stack</td>
771 <tr><td style="border: 2px solid blue">ROT</td>
772 <td style="border: 2px solid blue">ROT</td>
773 <td style="border: 2px solid blue">w1 w2 w3 -- w2 w3 w1</td>
774 <td style="border: 2px solid blue">The top three values are rotated. That is, three value are popped
775 off the stack. They are pushed back onto the stack in the order
778 <tr><td style="border: 2px solid blue">ROT2</td>
779 <td style="border: 2px solid blue">ROT2</td>
780 <td style="border: 2px solid blue">w1 w2 w3 w4 w5 w6 -- w3 w4 w5 w6 w1 w2</td>
781 <td style="border: 2px solid blue">Like ROT but the rotation is done using three pairs instead of
784 <tr><td style="border: 2px solid blue">RROT</td>
785 <td style="border: 2px solid blue">RROT</td>
786 <td style="border: 2px solid blue">w1 w2 w3 -- w2 w3 w1</td>
787 <td style="border: 2px solid blue">Reverse rotation. Like ROT, but it rotates the other way around.
788 Essentially, the third element on the stack is moved to the top
791 <tr><td style="border: 2px solid blue">RROT2</td>
792 <td style="border: 2px solid blue">RROT2</td>
793 <td style="border: 2px solid blue">w1 w2 w3 w4 w5 w6 -- w3 w4 w5 w6 w1 w2</td>
794 <td style="border: 2px solid blue">Double reverse rotation. Like RROT but the rotation is done using
795 three pairs instead of three singles. The fifth and sixth stack
796 elements are moved to the first and second positions</td>
798 <tr><td style="border: 2px solid blue">TUCK</td>
799 <td style="border: 2px solid blue">TUCK</td>
800 <td style="border: 2px solid blue">w1 w2 -- w2 w1 w2</td>
801 <td style="border: 2px solid blue">Similar to OVER except that the second operand is being
802 replicated. Essentially, the first operand is being "tucked"
803 in between two instances of the second operand. Logically, two
804 values are popped off the stack. They are placed back on the
805 stack in the order w2 w1 w2.</td>
807 <tr><td style="border: 2px solid blue">TUCK2</td>
808 <td style="border: 2px solid blue">TUCK2</td>
809 <td style="border: 2px solid blue">w1 w2 w3 w4 -- w3 w4 w1 w2 w3 w4</td>
810 <td style="border: 2px solid blue">Like TUCK but a pair of elements is tucked over two pairs.
811 That is, the top two elements of the stack are duplicated and
812 inserted into the stack at the fifth and positions.</td>
814 <tr><td style="border: 2px solid blue">PICK</td>
815 <td style="border: 2px solid blue">PICK</td>
816 <td style="border: 2px solid blue">x0 ... Xn n -- x0 ... Xn x0</td>
817 <td style="border: 2px solid blue">The top of the stack is used as an index into the remainder of
818 the stack. The element at the nth position replaces the index
819 (top of stack). This is useful for cycling through a set of
820 values. Note that indexing is zero based. So, if n=0 then you
821 get the second item on the stack. If n=1 you get the third, etc.
822 Note also that the index is replaced by the n'th value. </td>
824 <tr><td style="border: 2px solid blue">SELECT</td>
825 <td style="border: 2px solid blue">SELECT</td>
826 <td style="border: 2px solid blue">m n X0..Xm Xm+1 .. Xn -- Xm</td>
827 <td style="border: 2px solid blue">This is like PICK but the list is removed and you need to specify
828 both the index and the size of the list. Careful with this one,
829 the wrong value for n can blow away a huge amount of the stack.</td>
831 <tr><td style="border: 2px solid blue">ROLL</td>
832 <td style="border: 2px solid blue">ROLL</td>
833 <td style="border: 2px solid blue">x0 x1 .. xn n -- x1 .. xn x0</td>
834 <td style="border: 2px solid blue"><b>Not Implemented</b>. This one has been left as an exercise to
835 the student. See <a href="#exercise">Exercise</a>. ROLL requires
836 a value, "n", to be on the top of the stack. This value specifies how
837 far into the stack to "roll". The n'th value is <em>moved</em> (not
838 copied) from its location and replaces the "n" value on the top of the
839 stack. In this way, all the values between "n" and x0 roll up the stack.
840 The operation of ROLL is a generalized ROT. The "n" value specifies
841 how much to rotate. That is, ROLL with n=1 is the same as ROT and
842 ROLL with n=2 is the same as ROT2.</td>
844 <tr><td colspan="4"><b>MEMORY OPERATORS</b></td></tr>
846 <td style="border: 2px solid blue"><u>Word</u></td>
847 <td style="border: 2px solid blue"><u>Name</u></td>
848 <td style="border: 2px solid blue"><u>Operation</u></td>
849 <td style="border: 2px solid blue"><u>Description</u></td>
851 <tr><td style="border: 2px solid blue">MALLOC</td>
852 <td style="border: 2px solid blue">MALLOC</td>
853 <td style="border: 2px solid blue">w1 -- p</td>
854 <td style="border: 2px solid blue">One value is popped off the stack. The value is used as the size
855 of a memory block to allocate. The size is in bytes, not words.
856 The memory allocation is completed and the address of the memory
857 block is pushed onto the stack.</td>
859 <tr><td style="border: 2px solid blue">FREE</td>
860 <td style="border: 2px solid blue">FREE</td>
861 <td style="border: 2px solid blue">p -- </td>
862 <td style="border: 2px solid blue">One pointer value is popped off the stack. The value should be
863 the address of a memory block created by the MALLOC operation. The
864 associated memory block is freed. Nothing is pushed back on the
865 stack. Many bugs can be created by attempting to FREE something
866 that isn't a pointer to a MALLOC allocated memory block. Make
867 sure you know what's on the stack. One way to do this is with
868 the following idiom:<br/>
869 <code>64 MALLOC DUP DUP (use ptr) DUP (use ptr) ... FREE</code>
870 <br/>This ensures that an extra copy of the pointer is placed on
871 the stack (for the FREE at the end) and that every use of the
872 pointer is preceded by a DUP to retain the copy for FREE.</td>
874 <tr><td style="border: 2px solid blue">GET</td>
875 <td style="border: 2px solid blue">GET</td>
876 <td style="border: 2px solid blue">w1 p -- w2 p</td>
877 <td style="border: 2px solid blue">An integer index and a pointer to a memory block are popped of
878 the block. The index is used to index one byte from the memory
879 block. That byte value is retained, the pointer is pushed again
880 and the retained value is pushed. Note that the pointer value
881 s essentially retained in its position so this doesn't count
882 as a "use ptr" in the FREE idiom.</td>
884 <tr><td style="border: 2px solid blue">PUT</td>
885 <td style="border: 2px solid blue">PUT</td>
886 <td style="border: 2px solid blue">w1 w2 p -- p </td>
887 <td style="border: 2px solid blue">An integer value is popped of the stack. This is the value to
888 be put into a memory block. Another integer value is popped of
889 the stack. This is the indexed byte in the memory block. A
890 pointer to the memory block is popped off the stack. The
891 first value (w1) is then converted to a byte and written
892 to the element of the memory block(p) at the index given
893 by the second value (w2). The pointer to the memory block is
894 pushed back on the stack so this doesn't count as a "use ptr"
895 in the FREE idiom.</td>
897 <tr><td colspan="4"><b>CONTROL FLOW OPERATORS</b></td></tr>
899 <td style="border: 2px solid blue"><u>Word</u></td>
900 <td style="border: 2px solid blue"><u>Name</u></td>
901 <td style="border: 2px solid blue"><u>Operation</u></td>
902 <td style="border: 2px solid blue"><u>Description</u></td>
904 <tr><td style="border: 2px solid blue">RETURN</td>
905 <td style="border: 2px solid blue">RETURN</td>
906 <td style="border: 2px solid blue"> -- </td>
907 <td style="border: 2px solid blue">The currently executing definition returns immediately to its caller.
908 Note that there is an implicit <code>RETURN</code> at the end of each
909 definition, logically located at the semi-colon. The sequence
910 <code>RETURN ;</code> is valid but redundant.</td>
912 <tr><td style="border: 2px solid blue">EXIT</td>
913 <td style="border: 2px solid blue">EXIT</td>
914 <td style="border: 2px solid blue">w1 -- </td>
915 <td style="border: 2px solid blue">A return value for the program is popped off the stack. The program is
916 then immediately terminated. This is normally an abnormal exit from the
917 program. For a normal exit (when <code>MAIN</code> finishes), the exit
918 code will always be zero in accordance with UNIX conventions.</td>
920 <tr><td style="border: 2px solid blue">RECURSE</td>
921 <td style="border: 2px solid blue">RECURSE</td>
922 <td style="border: 2px solid blue"> -- </td>
923 <td style="border: 2px solid blue">The currently executed definition is called again. This operation is
924 needed since the definition of a word doesn't exist until the semi colon
925 is reacher. Attempting something like:<br/>
926 <code> : recurser recurser ; </code><br/> will yield and error saying that
927 "recurser" is not defined yet. To accomplish the same thing, change this
929 <code> : recurser RECURSE ; </code></td>
931 <tr><td style="border: 2px solid blue">IF (words...) ENDIF</td>
932 <td style="border: 2px solid blue">IF (words...) ENDIF</td>
933 <td style="border: 2px solid blue">b -- </td>
934 <td style="border: 2px solid blue">A boolean value is popped of the stack. If it is non-zero then the "words..."
935 are executed. Otherwise, execution continues immediately following the ENDIF.</td>
937 <tr><td style="border: 2px solid blue">IF (words...) ELSE (words...) ENDIF</td>
938 <td style="border: 2px solid blue">IF (words...) ELSE (words...) ENDIF</td>
939 <td style="border: 2px solid blue">b -- </td>
940 <td style="border: 2px solid blue">A boolean value is popped of the stack. If it is non-zero then the "words..."
941 between IF and ELSE are executed. Otherwise the words between ELSE and ENDIF are
942 executed. In either case, after the (words....) have executed, execution continues
943 immediately following the ENDIF. </td>
945 <tr><td style="border: 2px solid blue">WHILE (words...) END</td>
946 <td style="border: 2px solid blue">WHILE (words...) END</td>
947 <td style="border: 2px solid blue">b -- b </td>
948 <td style="border: 2px solid blue">The boolean value on the top of the stack is examined. If it is non-zero then the
949 "words..." between WHILE and END are executed. Execution then begins again at the WHILE where another
950 boolean is popped off the stack. To prevent this operation from eating up the entire
951 stack, you should push onto the stack (just before the END) a boolean value that indicates
952 whether to terminate. Note that since booleans and integers can be coerced you can
953 use the following "for loop" idiom:<br/>
954 <code>(push count) WHILE (words...) -- END</code><br/>
956 <code>10 WHILE DUP >d -- END</code><br/>
957 This will print the numbers from 10 down to 1. 10 is pushed on the stack. Since that is
958 non-zero, the while loop is entered. The top of the stack (10) is duplicated and then
959 printed out with >d. The top of the stack is decremented, yielding 9 and control is
960 transfered back to the WHILE keyword. The process starts all over again and repeats until
961 the top of stack is decremented to 0 at which the WHILE test fails and control is
962 transfered to the word after the END.</td>
964 <tr><td colspan="4"><b>INPUT & OUTPUT OPERATORS</b></td></tr>
966 <td style="border: 2px solid blue"><u>Word</u></td>
967 <td style="border: 2px solid blue"><u>Name</u></td>
968 <td style="border: 2px solid blue"><u>Operation</u></td>
969 <td style="border: 2px solid blue"><u>Description</u></td>
971 <tr><td style="border: 2px solid blue">SPACE</td>
972 <td style="border: 2px solid blue">SPACE</td>
973 <td style="border: 2px solid blue"> -- </td>
974 <td style="border: 2px solid blue">A space character is put out. There is no stack effect.</td>
976 <tr><td style="border: 2px solid blue">TAB</td>
977 <td style="border: 2px solid blue">TAB</td>
978 <td style="border: 2px solid blue"> -- </td>
979 <td style="border: 2px solid blue">A tab character is put out. There is no stack effect.</td>
981 <tr><td style="border: 2px solid blue">CR</td>
982 <td style="border: 2px solid blue">CR</td>
983 <td style="border: 2px solid blue"> -- </td>
984 <td style="border: 2px solid blue">A carriage return character is put out. There is no stack effect.</td>
986 <tr><td style="border: 2px solid blue">>s</td>
987 <td style="border: 2px solid blue">OUT_STR</td>
988 <td style="border: 2px solid blue"> -- </td>
989 <td style="border: 2px solid blue">A string pointer is popped from the stack. It is put out.</td>
991 <tr><td style="border: 2px solid blue">>d</td>
992 <td style="border: 2px solid blue">OUT_STR</td>
993 <td style="border: 2px solid blue"> -- </td>
994 <td style="border: 2px solid blue">A value is popped from the stack. It is put out as a decimal integer.</td>
996 <tr><td style="border: 2px solid blue">>c</td>
997 <td style="border: 2px solid blue">OUT_CHR</td>
998 <td style="border: 2px solid blue"> -- </td>
999 <td style="border: 2px solid blue">A value is popped from the stack. It is put out as an ASCII character.</td>
1001 <tr><td style="border: 2px solid blue"><s</td>
1002 <td style="border: 2px solid blue">IN_STR</td>
1003 <td style="border: 2px solid blue"> -- s </td>
1004 <td style="border: 2px solid blue">A string is read from the input via the scanf(3) format string " %as". The
1005 resulting string is pushed onto the stack.</td>
1007 <tr><td style="border: 2px solid blue"><d</td>
1008 <td style="border: 2px solid blue">IN_STR</td>
1009 <td style="border: 2px solid blue"> -- w </td>
1010 <td style="border: 2px solid blue">An integer is read from the input via the scanf(3) format string " %d". The
1011 resulting value is pushed onto the stack</td>
1013 <tr><td style="border: 2px solid blue"><c</td>
1014 <td style="border: 2px solid blue">IN_CHR</td>
1015 <td style="border: 2px solid blue"> -- w </td>
1016 <td style="border: 2px solid blue">A single character is read from the input via the scanf(3) format string
1017 " %c". The value is converted to an integer and pushed onto the stack.</td>
1019 <tr><td style="border: 2px solid blue">DUMP</td>
1020 <td style="border: 2px solid blue">DUMP</td>
1021 <td style="border: 2px solid blue"> -- </td>
1022 <td style="border: 2px solid blue">The stack contents are dumped to standard output. This is useful for
1023 debugging your definitions. Put DUMP at the beginning and end of a definition
1024 to see instantly the net effect of the definition.</td>
1028 <!-- ======================================================================= -->
1029 <div class="doc_section"> <a name="example">Prime: A Complete Example</a></div>
1030 <div class="doc_text">
1031 <p>The following fully documented program highlights many features of both
1032 the Stacker language and what is possible with LLVM. The program has two modes
1033 of operations. If you provide numeric arguments to the program, it checks to see
1034 if those arguments are prime numbers and prints out the results. Without any
1035 aruments, the program prints out any prime numbers it finds between 1 and one
1036 million (there's a lot of them!). The source code comments below tell the
1037 remainder of the story.
1040 <div class="doc_text">
1042 ################################################################################
1044 # Brute force prime number generator
1046 # This program is written in classic Stacker style, that being the style of a
1047 # stack. Start at the bottom and read your way up !
1049 # Reid Spencer - Nov 2003
1050 ################################################################################
1051 # Utility definitions
1052 ################################################################################
1054 : it_is_a_prime TRUE ;
1055 : it_is_not_a_prime FALSE ;
1056 : continue_loop TRUE ;
1059 ################################################################################
1060 # This definition tryies an actual division of a candidate prime number. It
1061 # determines whether the division loop on this candidate should continue or
1064 # div - the divisor to try
1065 # p - the prime number we are working on
1067 # cont - should we continue the loop ?
1068 # div - the next divisor to try
1069 # p - the prime number we are working on
1070 ################################################################################
1072 DUP2 ( save div and p )
1073 SWAP ( swap to put divisor second on stack)
1074 MOD 0 = ( get remainder after division and test for 0 )
1076 exit_loop ( remainder = 0, time to exit )
1078 continue_loop ( remainder != 0, keep going )
1082 ################################################################################
1083 # This function tries one divisor by calling try_dividing. But, before doing
1084 # that it checks to see if the value is 1. If it is, it does not bother with
1085 # the division because prime numbers are allowed to be divided by one. The
1086 # top stack value (cont) is set to determine if the loop should continue on
1087 # this prime number or not.
1089 # cont - should we continue the loop (ignored)?
1090 # div - the divisor to try
1091 # p - the prime number we are working on
1093 # cont - should we continue the loop ?
1094 # div - the next divisor to try
1095 # p - the prime number we are working on
1096 ################################################################################
1098 DROP ( drop the loop continuation )
1099 DUP ( save the divisor )
1100 1 = IF ( see if divisor is == 1 )
1101 exit_loop ( no point dividing by 1 )
1103 try_dividing ( have to keep going )
1105 SWAP ( get divisor on top )
1107 SWAP ( put loop continuation back on top )
1110 ################################################################################
1111 # The number on the stack (p) is a candidate prime number that we must test to
1112 # determine if it really is a prime number. To do this, we divide it by every
1113 # number from one p-1 to 1. The division is handled in the try_one_divisor
1114 # definition which returns a loop continuation value (which we also seed with
1115 # the value 1). After the loop, we check the divisor. If it decremented all
1116 # the way to zero then we found a prime, otherwise we did not find one.
1118 # p - the prime number to check
1120 # yn - boolean indiating if its a prime or not
1121 # p - the prime number checked
1122 ################################################################################
1124 DUP ( duplicate to get divisor value ) )
1125 -- ( first divisor is one less than p )
1126 1 ( continue the loop )
1128 try_one_divisor ( see if its prime )
1130 DROP ( drop the continuation value )
1131 0 = IF ( test for divisor == 1 )
1132 it_is_a_prime ( we found one )
1134 it_is_not_a_prime ( nope, this one is not a prime )
1138 ################################################################################
1139 # This definition determines if the number on the top of the stack is a prime
1140 # or not. It does this by testing if the value is degenerate (<= 3) and
1141 # responding with yes, its a prime. Otherwise, it calls try_harder to actually
1142 # make some calculations to determine its primeness.
1144 # p - the prime number to check
1146 # yn - boolean indicating if its a prime or not
1147 # p - the prime number checked
1148 ################################################################################
1150 DUP ( save the prime number )
1151 3 >= IF ( see if its <= 3 )
1152 it_is_a_prime ( its <= 3 just indicate its prime )
1154 try_harder ( have to do a little more work )
1158 ################################################################################
1159 # This definition is called when it is time to exit the program, after we have
1160 # found a sufficiently large number of primes.
1163 ################################################################################
1165 "Finished" >s CR ( say we are finished )
1166 0 EXIT ( exit nicely )
1169 ################################################################################
1170 # This definition checks to see if the candidate is greater than the limit. If
1171 # it is, it terminates the program by calling done. Otherwise, it increments
1172 # the value and calls is_prime to determine if the candidate is a prime or not.
1173 # If it is a prime, it prints it. Note that the boolean result from is_prime is
1174 # gobbled by the following IF which returns the stack to just contining the
1175 # prime number just considered.
1177 # p - one less than the prime number to consider
1179 # p+1 - the prime number considered
1180 ################################################################################
1182 DUP ( save the prime number to consider )
1183 1000000 < IF ( check to see if we are done yet )
1184 done ( we are done, call "done" )
1186 ++ ( increment to next prime number )
1187 is_prime ( see if it is a prime )
1189 print ( it is, print it )
1193 ################################################################################
1194 # This definition starts at one, prints it out and continues into a loop calling
1195 # consider_prime on each iteration. The prime number candidate we are looking at
1196 # is incremented by consider_prime.
1199 ################################################################################
1201 "Prime Numbers: " >s CR ( say hello )
1202 DROP ( get rid of that pesky string )
1203 1 ( stoke the fires )
1204 print ( print the first one, we know its prime )
1205 WHILE ( loop while the prime to consider is non zero )
1206 consider_prime ( consider one prime number )
1210 ################################################################################
1212 ################################################################################
1214 >d ( Print the prime number )
1215 " is prime." ( push string to output )
1217 CR ( print carriage return )
1222 >d ( Print the prime number )
1223 " is NOT prime." ( push string to put out )
1224 >s ( put out the string )
1225 CR ( print carriage return )
1229 ################################################################################
1230 # This definition processes a single command line argument and determines if it
1231 # is a prime number or not.
1233 # n - number of arguments
1234 # arg1 - the prime numbers to examine
1236 # n-1 - one less than number of arguments
1237 # arg2 - we processed one argument
1238 ################################################################################
1240 -- ( decrement loop counter )
1241 SWAP ( get the argument value )
1242 is_prime IF ( determine if its prime )
1247 DROP ( done with that argument )
1250 ################################################################################
1251 # The MAIN program just prints a banner and processes its arguments.
1253 # n - number of arguments
1254 # ... - the arguments
1255 ################################################################################
1257 WHILE ( while there are more arguments )
1258 do_one_argument ( process one argument )
1262 ################################################################################
1263 # The MAIN program just prints a banner and processes its arguments.
1265 ################################################################################
1267 NIP ( get rid of the program name )
1268 -- ( reduce number of arguments )
1269 DUP ( save the arg counter )
1270 1 <= IF ( See if we got an argument )
1271 process_arguments ( tell user if they are prime )
1273 find_primes ( see how many we can find )
1275 0 ( push return code )
1280 <!-- ======================================================================= -->
1281 <div class="doc_section"> <a name="internal">Internals</a></div>
1282 <div class="doc_text">
1283 <p><b>This section is under construction.</b>
1284 <p>In the mean time, you can always read the code! It has comments!</p>
1286 <!-- ======================================================================= -->
1287 <div class="doc_subsection"> <a name="directory">Directory Structure</a></div>
1288 <div class="doc_text">
1289 <p>The source code, test programs, and sample programs can all be found
1290 under the LLVM "projects" directory. You will need to obtain the LLVM sources
1291 to find it (either via anonymous CVS or a tarball. See the
1292 <a href="GettingStarted.html">Getting Started</a> document).</p>
1293 <p>Under the "projects" directory there is a directory named "stacker". That
1294 directory contains everything, as follows:</p>
1296 <li><em>lib</em> - contains most of the source code
1298 <li><em>lib/compiler</em> - contains the compiler library
1299 <li><em>lib/runtime</em> - contains the runtime library
1301 <li><em>test</em> - contains the test programs</li>
1302 <li><em>tools</em> - contains the Stacker compiler main program, stkrc
1304 <li><em>lib/stkrc</em> - contains the Stacker compiler main program
1306 <li><em>sample</em> - contains the sample programs</li>
1309 <!-- ======================================================================= -->
1310 <div class="doc_subsection"><a name="lexer"></a>The Lexer</div>
1311 <div class="doc_text">
1312 <p>See projects/Stacker/lib/compiler/Lexer.l</p>
1314 <!-- ======================================================================= -->
1315 <div class="doc_subsection"><a name="parser"></a>The Parser</div>
1316 <div class="doc_text">
1317 <p>See projects/Stacker/lib/compiler/StackerParser.y</p>
1319 <!-- ======================================================================= -->
1320 <div class="doc_subsection"><a name="compiler"></a>The Compiler</div>
1321 <div class="doc_text">
1322 <p>See projects/Stacker/lib/compiler/StackerCompiler.cpp</p>
1324 <!-- ======================================================================= -->
1325 <div class="doc_subsection"><a name="runtime"></a>The Runtime</div>
1326 <div class="doc_text">
1327 <p>See projects/Stacker/lib/runtime/stacker_rt.c</p>
1329 <!-- ======================================================================= -->
1330 <div class="doc_subsection"><a name="driver"></a>Compiler Driver</div>
1331 <div class="doc_text">
1332 <p>See projects/Stacker/tools/stkrc/stkrc.cpp</p>
1334 <!-- ======================================================================= -->
1335 <div class="doc_subsection"><a name="tests"></a>Test Programs</div>
1336 <div class="doc_text">
1337 <p>See projects/Stacker/test/*.st</p>
1339 <!-- ======================================================================= -->
1340 <div class="doc_subsection"> <a name="exercise">Exercise</a></div>
1341 <div class="doc_text">
1342 <p>As you may have noted from a careful inspection of the Built-In word
1343 definitions, the ROLL word is not implemented. This word was left out of
1344 Stacker on purpose so that it can be an exercise for the student. The exercise
1345 is to implement the ROLL functionality (in your own workspace) and build a test
1346 program for it. If you can implement ROLL you understand Stacker and probably
1347 a fair amount about LLVM since this is one of the more complicated Stacker
1348 operations. The work will almost be completely limited to the
1349 <a href="#compiler">compiler</a>.
1350 <p>The ROLL word is already recognized by both the lexer and parser but ignored
1351 by the compiler. That means you don't have to futz around with figuring out how
1352 to get the keyword recognized. It already is. The part of the compiler that
1353 you need to implement is the <code>ROLL</code> case in the
1354 <code>StackerCompiler::handle_word(int)</code> method.</p> See the implementations
1355 of PICk and SELECT in the same method to get some hints about how to complete
1359 <!-- ======================================================================= -->
1360 <div class="doc_subsection"> <a name="todo">Things Remaining To Be Done</a></div>
1361 <div class="doc_text">
1362 <p>The initial implementation of Stacker has several deficiencies. If you're
1363 interested, here are some things that could be implemented better:</p>
1365 <li>Write an LLVM pass to compute the correct stack depth needed by the
1366 program. Currently the stack is set to a fixed number which means programs
1367 with large numbers of definitions might fail.</li>
1368 <li>Enhance to run on 64-bit platforms like SPARC. Right now the size of a
1369 pointer on 64-bit machines will cause incorrect results because of the 32-bit
1370 size of a stack element currently supported. This feature was not implemented
1371 because LLVM needs a union type to be able to support the different sizes
1372 correctly (portably and efficiently).</li>
1373 <li>Write an LLVM pass to optimize the use of the global stack. The code
1374 emitted currently is somewhat wasteful. It gets cleaned up a lot by existing
1375 passes but more could be done.</li>
1376 <li>Add -O -O1 -O2 and -O3 optimization switches to the compiler driver to
1377 allow LLVM optimization without using "opt"</li>
1378 <li>Make the compiler driver use the LLVM linking facilities (with IPO) before
1379 depending on GCC to do the final link.</li>
1380 <li>Clean up parsing. It doesn't handle errors very well.</li>
1381 <li>Rearrange the StackerCompiler.cpp code to make better use of inserting
1382 instructions before a block's terminating instruction. I didn't figure this
1383 technique out until I was nearly done with LLVM. As it is, its a bad example
1384 of how to insert instructions!</li>
1385 <li>Provide for I/O to arbitrary files instead of just stdin/stdout.</li>
1386 <li>Write additional built-in words; with inspiration from FORTH</li>
1387 <li>Write additional sample Stacker programs.</li>
1388 <li>Add your own compiler writing experiences and tips in the
1389 <a href="#lessons">Lessons I Learned About LLVM</a> section.</li>
1392 <!-- ======================================================================= -->
1394 <div class="doc_footer">
1395 <address><a href="mailto:rspencer@x10sys.com">Reid Spencer</a></address>
1396 <a href="http://llvm.cs.uiuc.edu">The LLVM Compiler Infrastructure</a>
1397 <br>Last modified: $Date$ </div>