1 /* A sometimes minimal FORTH compiler and tutorial for Linux / i386 systems. -*- asm -*-
2 By Richard W.M. Jones <rich@annexia.org> http://annexia.org/forth
3 This is PUBLIC DOMAIN (see public domain release statement below).
4 $Id: jonesforth.S,v 1.21 2007-09-23 11:00:30 rich Exp $
6 gcc -m32 -nostdlib -static -Wl,-Ttext,0 -o jonesforth jonesforth.S
10 INTRODUCTION ----------------------------------------------------------------------
12 FORTH is one of those alien languages which most working programmers regard in the same
13 way as Haskell, LISP, and so on. Something so strange that they'd rather any thoughts
14 of it just go away so they can get on with writing this paying code. But that's wrong
15 and if you care at all about programming then you should at least understand all these
16 languages, even if you will never use them.
18 LISP is the ultimate high-level language, and features from LISP are being added every
19 decade to the more common languages. But FORTH is in some ways the ultimate in low level
20 programming. Out of the box it lacks features like dynamic memory management and even
21 strings. In fact, at its primitive level it lacks even basic concepts like IF-statements
24 Why then would you want to learn FORTH? There are several very good reasons. First
25 and foremost, FORTH is minimal. You really can write a complete FORTH in, say, 2000
26 lines of code. I don't just mean a FORTH program, I mean a complete FORTH operating
27 system, environment and language. You could boot such a FORTH on a bare PC and it would
28 come up with a prompt where you could start doing useful work. The FORTH you have here
29 isn't minimal and uses a Linux process as its 'base PC' (both for the purposes of making
30 it a good tutorial). It's possible to completely understand the system. Who can say they
31 completely understand how Linux works, or gcc?
33 Secondly FORTH has a peculiar bootstrapping property. By that I mean that after writing
34 a little bit of assembly to talk to the hardware and implement a few primitives, all the
35 rest of the language and compiler is written in FORTH itself. Remember I said before
36 that FORTH lacked IF-statements and loops? Well of course it doesn't really because
37 such a lanuage would be useless, but my point was rather that IF-statements and loops are
38 written in FORTH itself.
40 Now of course this is common in other languages as well, and in those languages we call
41 them 'libraries'. For example in C, 'printf' is a library function written in C. But
42 in FORTH this goes way beyond mere libraries. Can you imagine writing C's 'if' in C?
43 And that brings me to my third reason: If you can write 'if' in FORTH, then why restrict
44 yourself to the usual if/while/for/switch constructs? You want a construct that iterates
45 over every other element in a list of numbers? You can add it to the language. What
46 about an operator which pulls in variables directly from a configuration file and makes
47 them available as FORTH variables? Or how about adding Makefile-like dependencies to
48 the language? No problem in FORTH. This concept isn't common in programming languages,
49 but it has a name (in fact two names): "macros" (by which I mean LISP-style macros, not
50 the lame C preprocessor) and "domain specific languages" (DSLs).
52 This tutorial isn't about learning FORTH as the language. I'll point you to some references
53 you should read if you're not familiar with using FORTH. This tutorial is about how to
54 write FORTH. In fact, until you understand how FORTH is written, you'll have only a very
55 superficial understanding of how to use it.
57 So if you're not familiar with FORTH or want to refresh your memory here are some online
60 http://en.wikipedia.org/wiki/Forth_%28programming_language%29
62 http://galileo.phys.virginia.edu/classes/551.jvn.fall01/primer.htm
64 http://wiki.laptop.org/go/Forth_Lessons
66 http://www.albany.net/~hello/simple.htm
68 Here is another "Why FORTH?" essay: http://www.jwdt.com/~paysan/why-forth.html
70 Discussion and criticism of this FORTH here: http://lambda-the-ultimate.org/node/2452
72 ACKNOWLEDGEMENTS ----------------------------------------------------------------------
74 This code draws heavily on the design of LINA FORTH (http://home.hccnet.nl/a.w.m.van.der.horst/lina.html)
75 by Albert van der Horst. Any similarities in the code are probably not accidental.
77 Also I used this document (http://ftp.funet.fi/pub/doc/IOCCC/1992/buzzard.2.design) which really
78 defies easy explanation.
80 PUBLIC DOMAIN ----------------------------------------------------------------------
82 I, the copyright holder of this work, hereby release it into the public domain. This applies worldwide.
84 In case this is not legally possible, I grant any entity the right to use this work for any purpose,
85 without any conditions, unless such conditions are required by law.
87 SETTING UP ----------------------------------------------------------------------
89 Let's get a few housekeeping things out of the way. Firstly because I need to draw lots of
90 ASCII-art diagrams to explain concepts, the best way to look at this is using a window which
91 uses a fixed width font and is at least this wide:
93 <------------------------------------------------------------------------------------------------------------------------>
95 Secondly make sure TABS are set to 8 characters. The following should be a vertical
96 line. If not, sort out your tabs.
102 Thirdly I assume that your screen is at least 50 characters high.
104 ASSEMBLING ----------------------------------------------------------------------
106 If you want to actually run this FORTH, rather than just read it, you will need Linux on an
107 i386. Linux because instead of programming directly to the hardware on a bare PC which I
108 could have done, I went for a simpler tutorial by assuming that the 'hardware' is a Linux
109 process with a few basic system calls (read, write and exit and that's about all). i386
110 is needed because I had to write the assembly for a processor, and i386 is by far the most
111 common. (Of course when I say 'i386', any 32- or 64-bit x86 processor will do. I'm compiling
112 this on a 64 bit AMD Opteron).
114 Again, to assemble this you will need gcc and gas (the GNU assembler). The commands to
115 assemble and run the code (save this file as 'jonesforth.S') are:
117 gcc -m32 -nostdlib -static -Wl,-Ttext,0 -o jonesforth jonesforth.S
120 You will see lots of 'Warning: unterminated string; newline inserted' messages from the
121 assembler. That's just because the GNU assembler doesn't have a good syntax for multi-line
122 strings (or rather it used to, but the developers removed it!) so I've abused the syntax
123 slightly to make things readable. Ignore these warnings.
125 If you want to run your own FORTH programs you can do:
127 ./jonesforth < myprog.f
129 If you want to load your own FORTH code and then continue reading user commands, you can do:
131 cat myfunctions.f - | ./jonesforth
133 ASSEMBLER ----------------------------------------------------------------------
135 (You can just skip to the next section -- you don't need to be able to read assembler to
136 follow this tutorial).
138 However if you do want to read the assembly code here are a few notes about gas (the GNU assembler):
140 (1) Register names are prefixed with '%', so %eax is the 32 bit i386 accumulator. The registers
141 available on i386 are: %eax, %ebx, %ecx, %edx, %esi, %edi, %ebp and %esp, and most of them
142 have special purposes.
144 (2) Add, mov, etc. take arguments in the form SRC,DEST. So mov %eax,%ecx moves %eax -> %ecx
146 (3) Constants are prefixed with '$', and you mustn't forget it! If you forget it then it
147 causes a read from memory instead, so:
148 mov $2,%eax moves number 2 into %eax
149 mov 2,%eax reads the 32 bit word from address 2 into %eax (ie. most likely a mistake)
151 (4) gas has a funky syntax for local labels, where '1f' (etc.) means label '1:' "forwards"
152 and '1b' (etc.) means label '1:' "backwards".
154 (5) 'ja' is "jump if above", 'jb' for "jump if below", 'je' "jump if equal" etc.
156 (6) gas has a reasonably nice .macro syntax, and I use them a lot to make the code shorter and
159 For more help reading the assembler, do "info gas" at the Linux prompt.
161 Now the tutorial starts in earnest.
163 THE DICTIONARY ----------------------------------------------------------------------
165 In FORTH as you will know, functions are called "words", and just as in other languages they
166 have a name and a definition. Here are two FORTH words:
168 : DOUBLE DUP + ; \ name is "DOUBLE", definition is "DUP +"
169 : QUADRUPLE DOUBLE DOUBLE ; \ name is "QUADRUPLE", definition is "DOUBLE DOUBLE"
171 Words, both built-in ones and ones which the programmer defines later, are stored in a dictionary
172 which is just a linked list of dictionary entries.
174 <--- DICTIONARY ENTRY (HEADER) ----------------------->
175 +------------------------+--------+---------- - - - - +----------- - - - -
176 | LINK POINTER | LENGTH/| NAME | DEFINITION
178 +--- (4 bytes) ----------+- byte -+- n bytes - - - - +----------- - - - -
180 I'll come to the definition of the word later. For now just look at the header. The first
181 4 bytes are the link pointer. This points back to the previous word in the dictionary, or, for
182 the first word in the dictionary it is just a NULL pointer. Then comes a length/flags byte.
183 The length of the word can be up to 31 characters (5 bits used) and the top three bits are used
184 for various flags which I'll come to later. This is followed by the name itself, and in this
185 implementation the name is rounded up to a multiple of 4 bytes by padding it with zero bytes.
186 That's just to ensure that the definition starts on a 32 bit boundary.
188 A FORTH variable called LATEST contains a pointer to the most recently defined word, in
189 other words, the head of this linked list.
191 DOUBLE and QUADRUPLE might look like this:
193 pointer to previous word
196 +--|------+---+---+---+---+---+---+---+---+------------- - - - -
197 | LINK | 6 | D | O | U | B | L | E | 0 | (definition ...)
198 +---------+---+---+---+---+---+---+---+---+------------- - - - -
201 +--|------+---+---+---+---+---+---+---+---+---+---+---+---+------------- - - - -
202 | LINK | 9 | Q | U | A | D | R | U | P | L | E | 0 | 0 | (definition ...)
203 +---------+---+---+---+---+---+---+---+---+---+---+---+---+------------- - - - -
209 You should be able to see from this how you might implement functions to find a word in
210 the dictionary (just walk along the dictionary entries starting at LATEST and matching
211 the names until you either find a match or hit the NULL pointer at the end of the dictionary);
212 and add a word to the dictionary (create a new definition, set its LINK to LATEST, and set
213 LATEST to point to the new word). We'll see precisely these functions implemented in
214 assembly code later on.
216 One interesting consequence of using a linked list is that you can redefine words, and
217 a newer definition of a word overrides an older one. This is an important concept in
218 FORTH because it means that any word (even "built-in" or "standard" words) can be
219 overridden with a new definition, either to enhance it, to make it faster or even to
220 disable it. However because of the way that FORTH words get compiled, which you'll
221 understand below, words defined using the old definition of a word continue to use
222 the old definition. Only words defined after the new definition use the new definition.
224 DIRECT THREADED CODE ----------------------------------------------------------------------
226 Now we'll get to the really crucial bit in understanding FORTH, so go and get a cup of tea
227 or coffee and settle down. It's fair to say that if you don't understand this section, then you
228 won't "get" how FORTH works, and that would be a failure on my part for not explaining it well.
229 So if after reading this section a few times you don't understand it, please email me
232 Let's talk first about what "threaded code" means. Imagine a peculiar version of C where
233 you are only allowed to call functions without arguments. (Don't worry for now that such a
234 language would be completely useless!) So in our peculiar C, code would look like this:
243 and so on. How would a function, say 'f' above, be compiled by a standard C compiler?
244 Probably into assembly code like this. On the right hand side I've written the actual
248 CALL a E8 08 00 00 00
249 CALL b E8 1C 00 00 00
250 CALL c E8 2C 00 00 00
251 ; ignore the return from the function for now
253 "E8" is the x86 machine code to "CALL" a function. In the first 20 years of computing
254 memory was hideously expensive and we might have worried about the wasted space being used
255 by the repeated "E8" bytes. We can save 20% in code size (and therefore, in expensive memory)
256 by compressing this into just:
258 08 00 00 00 Just the function addresses, without
259 1C 00 00 00 the CALL prefix.
262 [Historical note: If the execution model that FORTH uses looks strange from the following
263 paragraphs, then it was motivated entirely by the need to save memory on early computers.
264 This code compression isn't so important now when our machines have more memory in their L1
265 caches than those early computers had in total, but the execution model still has some
268 Of course this code won't run directly any more. Instead we need to write an interpreter
269 which takes each pair of bytes and calls it.
271 On an i386 machine it turns out that we can write this interpreter rather easily, in just
272 two assembly instructions which turn into just 3 bytes of machine code. Let's store the
273 pointer to the next word to execute in the %esi register:
275 08 00 00 00 <- We're executing this one now. %esi is the _next_ one to execute.
279 The all-important i386 instruction is called LODSL (or in Intel manuals, LODSW). It does
280 two things. Firstly it reads the memory at %esi into the accumulator (%eax). Secondly it
281 increments %esi by 4 bytes. So after LODSL, the situation now looks like this:
283 08 00 00 00 <- We're still executing this one
284 1C 00 00 00 <- %eax now contains this address (0x0000001C)
287 Now we just need to jump to the address in %eax. This is again just a single x86 instruction
288 written JMP *(%eax). And after doing the jump, the situation looks like:
291 1C 00 00 00 <- Now we're executing this subroutine.
294 To make this work, each subroutine is followed by the two instructions 'LODSL; JMP *(%eax)'
295 which literally make the jump to the next subroutine.
297 And that brings us to our first piece of actual code! Well, it's a macro.
306 /* The macro is called NEXT. That's a FORTH-ism. It expands to those two instructions.
308 Every FORTH primitive that we write has to be ended by NEXT. Think of it kind of like
311 The above describes what is known as direct threaded code.
313 To sum up: We compress our function calls down to a list of addresses and use a somewhat
314 magical macro to act as a "jump to next function in the list". We also use one register (%esi)
315 to act as a kind of instruction pointer, pointing to the next function in the list.
317 I'll just give you a hint of what is to come by saying that a FORTH definition such as:
319 : QUADRUPLE DOUBLE DOUBLE ;
321 actually compiles (almost, not precisely but we'll see why in a moment) to a list of
322 function addresses for DOUBLE, DOUBLE and a special function called EXIT to finish off.
324 At this point, REALLY EAGLE-EYED ASSEMBLY EXPERTS are saying "JONES, YOU'VE MADE A MISTAKE!".
326 I lied about JMP *(%eax).
328 INDIRECT THREADED CODE ----------------------------------------------------------------------
330 It turns out that direct threaded code is interesting but only if you want to just execute
331 a list of functions written in assembly language. So QUADRUPLE would work only if DOUBLE
332 was an assembly language function. In the direct threaded code, QUADRUPLE would look like:
335 | addr of DOUBLE --------------------> (assembly code to do the double)
336 +------------------+ NEXT
337 %esi -> | addr of DOUBLE |
340 We can add an extra indirection to allow us to run both words written in assembly language
341 (primitives written for speed) and words written in FORTH themselves as lists of addresses.
343 The extra indirection is the reason for the brackets in JMP *(%eax).
345 Let's have a look at how QUADRUPLE and DOUBLE really look in FORTH:
347 : QUADRUPLE DOUBLE DOUBLE ;
350 | codeword | : DOUBLE DUP + ;
352 | addr of DOUBLE ---------------> +------------------+
353 +------------------+ | codeword |
354 | addr of DOUBLE | +------------------+
355 +------------------+ | addr of DUP --------------> +------------------+
356 | addr of EXIT | +------------------+ | codeword -------+
357 +------------------+ %esi -> | addr of + --------+ +------------------+ |
358 +------------------+ | | assembly to <-----+
359 | addr of EXIT | | | implement DUP |
360 +------------------+ | | .. |
363 | +------------------+
365 +-----> +------------------+
367 +------------------+ |
368 | assembly to <------+
375 This is the part where you may need an extra cup of tea/coffee/favourite caffeinated
376 beverage. What has changed is that I've added an extra pointer to the beginning of
377 the definitions. In FORTH this is sometimes called the "codeword". The codeword is
378 a pointer to the interpreter to run the function. For primitives written in
379 assembly language, the "interpreter" just points to the actual assembly code itself.
380 They don't need interpreting, they just run.
382 In words written in FORTH (like QUADRUPLE and DOUBLE), the codeword points to an interpreter
385 I'll show you the interpreter function shortly, but let's recall our indirect
386 JMP *(%eax) with the "extra" brackets. Take the case where we're executing DOUBLE
387 as shown, and DUP has been called. Note that %esi is pointing to the address of +
389 The assembly code for DUP eventually does a NEXT. That:
391 (1) reads the address of + into %eax %eax points to the codeword of +
392 (2) increments %esi by 4
393 (3) jumps to the indirect %eax jumps to the address in the codeword of +,
394 ie. the assembly code to implement +
399 | addr of DOUBLE ---------------> +------------------+
400 +------------------+ | codeword |
401 | addr of DOUBLE | +------------------+
402 +------------------+ | addr of DUP --------------> +------------------+
403 | addr of EXIT | +------------------+ | codeword -------+
404 +------------------+ | addr of + --------+ +------------------+ |
405 +------------------+ | | assembly to <-----+
406 %esi -> | addr of EXIT | | | implement DUP |
407 +------------------+ | | .. |
410 | +------------------+
412 +-----> +------------------+
414 +------------------+ |
415 now we're | assembly to <-----+
416 executing | implement + |
422 So I hope that I've convinced you that NEXT does roughly what you'd expect. This is
423 indirect threaded code.
425 I've glossed over four things. I wonder if you can guess without reading on what they are?
431 My list of four things are: (1) What does "EXIT" do? (2) which is related to (1) is how do
432 you call into a function, ie. how does %esi start off pointing at part of QUADRUPLE, but
433 then point at part of DOUBLE. (3) What goes in the codeword for the words which are written
434 in FORTH? (4) How do you compile a function which does anything except call other functions
435 ie. a function which contains a number like : DOUBLE 2 * ; ?
437 THE INTERPRETER AND RETURN STACK ------------------------------------------------------------
439 Going at these in no particular order, let's talk about issues (3) and (2), the interpreter
440 and the return stack.
442 Words which are defined in FORTH need a codeword which points to a little bit of code to
443 give them a "helping hand" in life. They don't need much, but they do need what is known
444 as an "interpreter", although it doesn't really "interpret" in the same way that, say,
445 Java bytecode used to be interpreted (ie. slowly). This interpreter just sets up a few
446 machine registers so that the word can then execute at full speed using the indirect
447 threaded model above.
449 One of the things that needs to happen when QUADRUPLE calls DOUBLE is that we save the old
450 %esi ("instruction pointer") and create a new one pointing to the first word in DOUBLE.
451 Because we will need to restore the old %esi at the end of DOUBLE (this is, after all, like
452 a function call), we will need a stack to store these "return addresses" (old values of %esi).
454 As you will have read, when reading the background documentation, FORTH has two stacks,
455 an ordinary stack for parameters, and a return stack which is a bit more mysterious. But
456 our return stack is just the stack I talked about in the previous paragraph, used to save
457 %esi when calling from a FORTH word into another FORTH word.
459 In this FORTH, we are using the normal stack pointer (%esp) for the parameter stack.
460 We will use the i386's "other" stack pointer (%ebp, usually called the "frame pointer")
461 for our return stack.
463 I've got two macros which just wrap up the details of using %ebp for the return stack.
464 You use them as for example "PUSHRSP %eax" (push %eax on the return stack) or "POPRSP %ebx"
465 (pop top of return stack into %ebx).
468 /* Macros to deal with the return stack. */
470 lea -4(%ebp),%ebp // push reg on to return stack
475 mov (%ebp),\reg // pop top of return stack to reg
480 And with that we can now talk about the interpreter.
482 In FORTH the interpreter function is often called DOCOL (I think it means "DO COLON" because
483 all FORTH definitions start with a colon, as in : DOUBLE DUP + ;
485 The "interpreter" (it's not really "interpreting") just needs to push the old %esi on the
486 stack and set %esi to the first word in the definition. Remember that we jumped to the
487 function using JMP *(%eax)? Well a consequence of that is that conveniently %eax contains
488 the address of this codeword, so just by adding 4 to it we get the address of the first
489 data word. Finally after setting up %esi, it just does NEXT which causes that first word
493 /* DOCOL - the interpreter! */
497 PUSHRSP %esi // push %esi on to the return stack
498 addl $4,%eax // %eax points to codeword, so make
499 movl %eax,%esi // %esi point to first data word
503 Just to make this absolutely clear, let's see how DOCOL works when jumping from QUADRUPLE
509 +------------------+ DOUBLE:
510 | addr of DOUBLE ---------------> +------------------+
511 +------------------+ %eax -> | addr of DOCOL |
512 %esi -> | addr of DOUBLE | +------------------+
513 +------------------+ | addr of DUP |
514 | addr of EXIT | +------------------+
515 +------------------+ | etc. |
517 First, the call to DOUBLE calls DOCOL (the codeword of DOUBLE). DOCOL does this: It
518 pushes the old %esi on the return stack. %eax points to the codeword of DOUBLE, so we
519 just add 4 on to it to get our new %esi:
524 +------------------+ DOUBLE:
525 | addr of DOUBLE ---------------> +------------------+
526 top of return +------------------+ %eax -> | addr of DOCOL |
527 stack points -> | addr of DOUBLE | + 4 = +------------------+
528 +------------------+ %esi -> | addr of DUP |
529 | addr of EXIT | +------------------+
530 +------------------+ | etc. |
532 Then we do NEXT, and because of the magic of threaded code that increments %esi again
535 Well, it seems to work.
537 One minor point here. Because DOCOL is the first bit of assembly actually to be defined
538 in this file (the others were just macros), and because I usually compile this code with the
539 text segment starting at address 0, DOCOL has address 0. So if you are disassembling the
540 code and see a word with a codeword of 0, you will immediately know that the word is
541 written in FORTH (it's not an assembler primitive) and so uses DOCOL as the interpreter.
543 STARTING UP ----------------------------------------------------------------------
545 Now let's get down to nuts and bolts. When we start the program we need to set up
546 a few things like the return stack. But as soon as we can, we want to jump into FORTH
547 code (albeit much of the "early" FORTH code will still need to be written as
548 assembly language primitives).
550 This is what the set up code does. Does a tiny bit of house-keeping, sets up the
551 separate return stack (NB: Linux gives us the ordinary parameter stack already), then
552 immediately jumps to a FORTH word called COLD. COLD stands for cold-start. In ISO
553 FORTH (but not in this FORTH), COLD can be called at any time to completely reset
554 the state of FORTH, and there is another word called WARM which does a partial reset.
557 /* ELF entry point. */
562 mov %esp,var_S0 // Store the initial data stack pointer.
563 mov $return_stack,%ebp // Initialise the return stack.
565 mov $cold_start,%esi // Initialise interpreter.
566 NEXT // Run interpreter!
569 cold_start: // High-level code without a codeword.
573 We also allocate some space for the return stack and some space to store user
574 definitions. These are static memory allocations using fixed-size buffers, but it
575 wouldn't be a great deal of work to make them dynamic.
579 /* FORTH return stack. */
580 #define RETURN_STACK_SIZE 8192
582 .space RETURN_STACK_SIZE
583 return_stack: // Initial top of return stack.
585 /* Space for user-defined words. */
586 #define USER_DEFS_SIZE 16384
589 .space USER_DEFS_SIZE
592 BUILT-IN WORDS ----------------------------------------------------------------------
594 Remember our dictionary entries (headers). Let's bring those together with the codeword
595 and data words to see how : DOUBLE DUP + ; really looks in memory.
597 pointer to previous word
600 +--|------+---+---+---+---+---+---+---+---+------------+------------+------------+------------+
601 | LINK | 6 | D | O | U | B | L | E | 0 | DOCOL | DUP | + | EXIT |
602 +---------+---+---+---+---+---+---+---+---+------------+--|---------+------------+------------+
605 LINK in next word points to codeword of DUP
607 Initially we can't just write ": DOUBLE DUP + ;" (ie. that literal string) here because we
608 don't yet have anything to read the string, break it up at spaces, parse each word, etc. etc.
609 So instead we will have to define built-in words using the GNU assembler data constructors
610 (like .int, .byte, .string, .ascii and so on -- look them up in the gas info page if you are
613 The long way would be:
614 .int <link to previous word>
616 .ascii "DOUBLE" // string
618 DOUBLE: .int DOCOL // codeword
619 .int DUP // pointer to codeword of DUP
620 .int PLUS // pointer to codeword of +
621 .int EXIT // pointer to codeword of EXIT
623 That's going to get quite tedious rather quickly, so here I define an assembler macro
624 so that I can just write:
626 defword "DOUBLE",6,,DOUBLE
629 and I'll get exactly the same effect.
631 Don't worry too much about the exact implementation details of this macro - it's complicated!
634 /* Flags - these are discussed later. */
636 #define F_HIDDEN 0x20
638 // Store the chain of links.
641 .macro defword name, namelen, flags=0, label
647 .set link,name_\label
648 .byte \flags+\namelen // flags + length byte
649 .ascii "\name" // the name
653 .int DOCOL // codeword - the interpreter
654 // list of word pointers follow
658 Similarly I want a way to write words written in assembly language. There will quite a few
659 of these to start with because, well, everything has to start in assembly before there's
660 enough "infrastructure" to be able to start writing FORTH words, but also I want to define
661 some common FORTH words in assembly language for speed, even though I could write them in FORTH.
663 This is what DUP looks like in memory:
665 pointer to previous word
668 +--|------+---+---+---+---+------------+
669 | LINK | 3 | D | U | P | code_DUP ---------------------> points to the assembly
670 +---------+---+---+---+---+------------+ code used to write DUP,
671 ^ len codeword which ends with NEXT.
675 Again, for brevity in writing the header I'm going to write an assembler macro called defcode.
678 .macro defcode name, namelen, flags=0, label
684 .set link,name_\label
685 .byte \flags+\namelen // flags + length byte
686 .ascii "\name" // the name
690 .int code_\label // codeword
694 code_\label : // assembler code follows
698 Now some easy FORTH primitives. These are written in assembly for speed. If you understand
699 i386 assembly language then it is worth reading these. However if you don't understand assembly
700 you can skip the details.
704 pop %eax // duplicate top of stack
709 defcode "DROP",4,,DROP
710 pop %eax // drop top of stack
713 defcode "SWAP",4,,SWAP
714 pop %eax // swap top of stack
720 defcode "OVER",4,,OVER
721 mov 4(%esp),%eax // get the second element of stack
722 push %eax // and push it on top
734 defcode "-ROT",4,,NROT
744 incl (%esp) // increment top of stack
748 decl (%esp) // decrement top of stack
751 defcode "4+",2,,INCR4
752 addl $4,(%esp) // add 4 to top of stack
755 defcode "4-",2,,DECR4
756 subl $4,(%esp) // subtract 4 from top of stack
760 pop %eax // get top of stack
761 addl %eax,(%esp) // and add it to next word on stack
765 pop %eax // get top of stack
766 subl %eax,(%esp) // and subtract it from next word on stack
773 push %eax // ignore overflow
781 push %eax // push quotient
789 push %edx // push remainder
792 defcode "=",1,,EQU // top two words are equal?
802 defcode "<>",2,,NEQU // top two words are not equal?
812 defcode "0=",2,,ZEQU // top of stack equals 0?
831 defcode "INVERT",6,,INVERT // this is the FORTH "NOT" function
836 RETURNING FROM FORTH WORDS ----------------------------------------------------------------------
838 Time to talk about what happens when we EXIT a function. In this diagram QUADRUPLE has called
839 DOUBLE, and DOUBLE is about to exit (look at where %esi is pointing):
844 +------------------+ DOUBLE
845 | addr of DOUBLE ---------------> +------------------+
846 +------------------+ | codeword |
847 | addr of DOUBLE | +------------------+
848 +------------------+ | addr of DUP |
849 | addr of EXIT | +------------------+
850 +------------------+ | addr of + |
852 %esi -> | addr of EXIT |
855 What happens when the + function does NEXT? Well, the following code is executed.
858 defcode "EXIT",4,,EXIT
859 POPRSP %esi // pop return stack into %esi
863 EXIT gets the old %esi which we saved from before on the return stack, and puts it in %esi.
864 So after this (but just before NEXT) we get:
869 +------------------+ DOUBLE
870 | addr of DOUBLE ---------------> +------------------+
871 +------------------+ | codeword |
872 %esi -> | addr of DOUBLE | +------------------+
873 +------------------+ | addr of DUP |
874 | addr of EXIT | +------------------+
875 +------------------+ | addr of + |
880 And NEXT just completes the job by, well in this case just by calling DOUBLE again :-)
882 LITERALS ----------------------------------------------------------------------
884 The final point I "glossed over" before was how to deal with functions that do anything
885 apart from calling other functions. For example, suppose that DOUBLE was defined like this:
889 It does the same thing, but how do we compile it since it contains the literal 2? One way
890 would be to have a function called "2" (which you'd have to write in assembler), but you'd need
891 a function for every single literal that you wanted to use.
893 FORTH solves this by compiling the function using a special word called LIT:
895 +---------------------------+-------+-------+-------+-------+-------+
896 | (usual header of DOUBLE) | DOCOL | LIT | 2 | * | EXIT |
897 +---------------------------+-------+-------+-------+-------+-------+
899 LIT is executed in the normal way, but what it does next is definitely not normal. It
900 looks at %esi (which now points to the literal 2), grabs it, pushes it on the stack, then
901 manipulates %esi in order to skip the literal as if it had never been there.
903 What's neat is that the whole grab/manipulate can be done using a single byte single
904 i386 instruction, our old friend LODSL. Rather than me drawing more ASCII-art diagrams,
905 see if you can find out how LIT works:
909 // %esi points to the next command, but in this case it points to the next
910 // literal 32 bit integer. Get that literal into %eax and increment %esi.
911 // On x86, it's a convenient single byte instruction! (cf. NEXT macro)
913 push %eax // push the literal number on to stack
917 MEMORY ----------------------------------------------------------------------
919 As important point about FORTH is that it gives you direct access to the lowest levels
920 of the machine. Manipulating memory directly is done frequently in FORTH, and these are
921 the primitive words for doing it.
925 pop %ebx // address to store at
926 pop %eax // data to store there
927 mov %eax,(%ebx) // store it
931 pop %ebx // address to fetch
932 mov (%ebx),%eax // fetch it
933 push %eax // push value onto stack
936 defcode "+!",2,,ADDSTORE
938 pop %eax // the amount to add
939 addl %eax,(%ebx) // add it
942 defcode "-!",2,,SUBSTORE
944 pop %eax // the amount to subtract
945 subl %eax,(%ebx) // add it
948 /* ! and @ (STORE and FETCH) store 32-bit words. It's also useful to be able to read and write bytes.
949 * I don't know whether FORTH has these words, so I invented my own, called !b and @b.
950 * Byte-oriented operations only work on architectures which permit them (i386 is one of those).
951 * UPDATE: writing a byte to the dictionary pointer is called C, in FORTH.
953 defcode "!b",2,,STOREBYTE
954 pop %ebx // address to store at
955 pop %eax // data to store there
956 movb %al,(%ebx) // store it
959 defcode "@b",2,,FETCHBYTE
960 pop %ebx // address to fetch
962 movb (%ebx),%al // fetch it
963 push %eax // push value onto stack
967 BUILT-IN VARIABLES ----------------------------------------------------------------------
969 These are some built-in variables and related standard FORTH words. Of these, the only one that we
970 have discussed so far was LATEST, which points to the last (most recently defined) word in the
971 FORTH dictionary. LATEST is also a FORTH word which pushes the address of LATEST (the variable)
972 on to the stack, so you can read or write it using @ and ! operators. For example, to print
973 the current value of LATEST (and this can apply to any FORTH variable) you would do:
977 To make defining variables shorter, I'm using a macro called defvar, similar to defword and
978 defcode above. (In fact the defvar macro uses defcode to do the dictionary header).
981 .macro defvar name, namelen, flags=0, label, initial=0
982 defcode \name,\namelen,\flags,\label
992 The built-in variables are:
994 STATE Is the interpreter executing code (0) or compiling a word (non-zero)?
995 LATEST Points to the latest (most recently defined) word in the dictionary.
996 HERE Points to the next free byte of memory. When compiling, compiled words go here.
997 _X These are three scratch variables, used by some standard dictionary words.
1000 S0 Stores the address of the top of the parameter stack.
1001 R0 Stores the address of the top of the return stack.
1002 VERSION Is the current version of this FORTH.
1005 defvar "STATE",5,,STATE
1006 defvar "HERE",4,,HERE,user_defs_start
1007 defvar "LATEST",6,,LATEST,name_SYSEXIT // SYSEXIT must be last in built-in dictionary
1012 defvar "R0",2,,RZ,return_stack
1013 defvar "VERSION",7,,VERSION,JONES_VERSION
1016 RETURN STACK ----------------------------------------------------------------------
1018 These words allow you to access the return stack. Recall that the register %ebp always points to
1019 the top of the return stack.
1023 pop %eax // pop parameter stack into %eax
1024 PUSHRSP %eax // push it on to the return stack
1027 defcode "R>",2,,FROMR
1028 POPRSP %eax // pop return stack on to %eax
1029 push %eax // and push on to parameter stack
1032 defcode "RSP@",4,,RSPFETCH
1036 defcode "RSP!",4,,RSPSTORE
1040 defcode "RDROP",5,,RDROP
1041 lea 4(%ebp),%ebp // pop return stack and throw away
1045 PARAMETER (DATA) STACK ----------------------------------------------------------------------
1047 These functions allow you to manipulate the parameter stack. Recall that Linux sets up the parameter
1048 stack for us, and it is accessed through %esp.
1051 defcode "DSP@",4,,DSPFETCH
1056 defcode "DSP!",4,,DSPSTORE
1061 INPUT AND OUTPUT ----------------------------------------------------------------------
1063 These are our first really meaty/complicated FORTH primitives. I have chosen to write them in
1064 assembler, but surprisingly in "real" FORTH implementations these are often written in terms
1065 of more fundamental FORTH primitives. I chose to avoid that because I think that just obscures
1066 the implementation. After all, you may not understand assembler but you can just think of it
1067 as an opaque block of code that does what it says.
1069 Let's discuss input first.
1071 The FORTH word KEY reads the next byte from stdin (and pushes it on the parameter stack).
1072 So if KEY is called and someone hits the space key, then the number 32 (ASCII code of space)
1073 is pushed on the stack.
1075 In FORTH there is no distinction between reading code and reading input. We might be reading
1076 and compiling code, we might be reading words to execute, we might be asking for the user
1077 to type their name -- ultimately it all comes in through KEY.
1079 The implementation of KEY uses an input buffer of a certain size (defined at the end of the
1080 program). It calls the Linux read(2) system call to fill this buffer and tracks its position
1081 in the buffer using a couple of variables, and if it runs out of input buffer then it refills
1082 it automatically. The other thing that KEY does is if it detects that stdin has closed, it
1083 exits the program, which is why when you hit ^D the FORTH system cleanly exits.
1086 #include <asm-i386/unistd.h>
1088 defcode "KEY",3,,KEY
1090 push %eax // push return value on stack
1102 1: // out of input; use read(2) to fetch more input from stdin
1103 xor %ebx,%ebx // 1st param: stdin
1104 mov $buffer,%ecx // 2nd param: buffer
1106 mov $buffend-buffer,%edx // 3rd param: max length
1107 mov $__NR_read,%eax // syscall: read
1109 test %eax,%eax // If %eax <= 0, then exit.
1111 addl %eax,%ecx // buffer+%eax = bufftop
1115 2: // error or out of input: exit
1117 mov $__NR_exit,%eax // syscall: exit
1121 By contrast, output is much simpler. The FORTH word EMIT writes out a single byte to stdout.
1122 This implementation just uses the write system call. No attempt is made to buffer output, but
1123 it would be a good exercise to add it.
1126 defcode "EMIT",4,,EMIT
1131 mov $1,%ebx // 1st param: stdout
1133 // write needs the address of the byte to write
1135 mov $2f,%ecx // 2nd param: address
1137 mov $1,%edx // 3rd param: nbytes = 1
1139 mov $__NR_write,%eax // write syscall
1144 2: .space 1 // scratch used by EMIT
1147 Back to input, WORD is a FORTH word which reads the next full word of input.
1149 What it does in detail is that it first skips any blanks (spaces, tabs, newlines and so on).
1150 Then it calls KEY to read characters into an internal buffer until it hits a blank. Then it
1151 calculates the length of the word it read and returns the address and the length as
1152 two words on the stack (with address at the top).
1154 Notice that WORD has a single internal buffer which it overwrites each time (rather like
1155 a static C string). Also notice that WORD's internal buffer is just 32 bytes long and
1156 there is NO checking for overflow. 31 bytes happens to be the maximum length of a
1157 FORTH word that we support, and that is what WORD is used for: to read FORTH words when
1158 we are compiling and executing code. The returned strings are not NUL-terminated, so
1159 in some crazy-world you could define FORTH words containing ASCII NULs, although why
1160 you'd want to is a bit beyond me.
1162 WORD is not suitable for just reading strings (eg. user input) because of all the above
1163 peculiarities and limitations.
1165 Note that when executing, you'll see:
1167 which puts "FOO" and length 3 on the stack, but when compiling:
1169 is an error (or at least it doesn't do what you might expect). Later we'll talk about compiling
1170 and immediate mode, and you'll understand why.
1173 defcode "WORD",4,,WORD
1175 push %ecx // push length
1176 push %edi // push base address
1180 /* Search for first non-blank character. Also skip \ comments. */
1182 call _KEY // get next key, returned in %eax
1183 cmpb $'\\',%al // start of a comment?
1184 je 3f // if so, skip the comment
1186 jbe 1b // if so, keep looking
1188 /* Search for the end of the word, storing chars as we go. */
1189 mov $5f,%edi // pointer to return buffer
1191 stosb // add character to return buffer
1192 call _KEY // get next key, returned in %al
1193 cmpb $' ',%al // is blank?
1194 ja 2b // if not, keep looping
1196 /* Return the word (well, the static buffer) and length. */
1198 mov %edi,%ecx // return length of the word
1199 mov $5f,%edi // return address of the word
1202 /* Code to skip \ comments to end of the current line. */
1205 cmpb $'\n',%al // end of line yet?
1210 // A static buffer where WORD returns. Subsequent calls
1211 // overwrite this buffer. Maximum word length is 32 chars.
1215 . (also called DOT) prints the top of the stack as an integer. In real FORTH implementations
1216 it should print it in the current base, but this assembler version is simpler and can only
1219 Remember that you can override even built-in FORTH words easily, so if you want to write a
1220 more advanced DOT then you can do so easily at a later point, and probably in FORTH.
1224 pop %eax // Get the number to print into %eax
1225 call _DOT // Easier to do this recursively ...
1228 mov $10,%ecx // Base 10
1232 xor %edx,%edx // %edx:%eax / %ecx -> quotient %eax, remainder %edx
1247 Almost the opposite of DOT (but not quite), SNUMBER parses a numeric string such as one returned
1248 by WORD and pushes the number on the parameter stack.
1250 This function does absolutely no error checking, and in particular the length of the string
1251 must be >= 1 bytes, and should contain only digits 0-9. If it doesn't you'll get random results.
1253 This function is only used when reading literal numbers in code, and shouldn't really be used
1254 in user code at all.
1256 defcode "SNUMBER",7,,SNUMBER
1266 imull $10,%eax // %eax *= 10
1269 subb $'0',%bl // ASCII -> digit
1276 DICTIONARY LOOK UPS ----------------------------------------------------------------------
1278 We're building up to our prelude on how FORTH code is compiled, but first we need yet more infrastructure.
1280 The FORTH word FIND takes a string (a word as parsed by WORD -- see above) and looks it up in the
1281 dictionary. What it actually returns is the address of the dictionary header, if it finds it,
1284 So if DOUBLE is defined in the dictionary, then WORD DOUBLE FIND returns the following pointer:
1290 +---------+---+---+---+---+---+---+---+---+------------+------------+------------+------------+
1291 | LINK | 6 | D | O | U | B | L | E | 0 | DOCOL | DUP | + | EXIT |
1292 +---------+---+---+---+---+---+---+---+---+------------+------------+------------+------------+
1294 See also >CFA which takes a dictionary entry pointer and returns a pointer to the codeword.
1296 FIND doesn't find dictionary entries which are flagged as HIDDEN. See below for why.
1299 defcode "FIND",4,,FIND
1300 pop %edi // %edi = address
1301 pop %ecx // %ecx = length
1307 push %esi // Save %esi so we can use it in string comparison.
1309 // Now we start searching backwards through the dictionary for this word.
1310 mov var_LATEST,%edx // LATEST points to name header of the latest word in the dictionary
1312 test %edx,%edx // NULL pointer? (end of the linked list)
1315 // Compare the length expected and the length of the word.
1316 // Note that if the F_HIDDEN flag is set on the word, then by a bit of trickery
1317 // this won't pick the word (the length will appear to be wrong).
1319 movb 4(%edx),%al // %al = flags+length field
1320 andb $(F_HIDDEN|0x1f),%al // %al = name length
1321 cmpb %cl,%al // Length is the same?
1324 // Compare the strings in detail.
1325 push %ecx // Save the length
1326 push %edi // Save the address (repe cmpsb will move this pointer)
1327 lea 5(%edx),%esi // Dictionary string we are checking against.
1328 repe cmpsb // Compare the strings.
1331 jne 2f // Not the same.
1333 // The strings are the same - return the header pointer in %eax
1339 mov (%edx),%edx // Move back through the link field to the previous word
1340 jmp 1b // .. and loop.
1344 xor %eax,%eax // Return zero to indicate not found.
1348 FIND returns the dictionary pointer, but when compiling we need the codeword pointer (recall
1349 that FORTH definitions are compiled into lists of codeword pointers). The standard FORTH
1350 word >CFA turns a dictionary pointer into a codeword pointer.
1352 The example below shows the result of:
1354 WORD DOUBLE FIND >CFA
1356 FIND returns a pointer to this
1357 | >CFA converts it to a pointer to this
1360 +---------+---+---+---+---+---+---+---+---+------------+------------+------------+------------+
1361 | LINK | 6 | D | O | U | B | L | E | 0 | DOCOL | DUP | + | EXIT |
1362 +---------+---+---+---+---+---+---+---+---+------------+------------+------------+------------+
1366 Because names vary in length, this isn't just a simple increment.
1368 In this FORTH you cannot easily turn a codeword pointer back into a dictionary entry pointer, but
1369 that is not true in most FORTH implementations where they store a back pointer in the definition
1370 (with an obvious memory/complexity cost). The reason they do this is that it is useful to be
1371 able to go backwards (codeword -> dictionary entry) in order to decompile FORTH definitions.
1373 What does CFA stand for? My best guess is "Code Field Address".
1376 defcode ">CFA",4,,TCFA
1383 add $4,%edi // Skip link pointer.
1384 movb (%edi),%al // Load flags+len into %al.
1385 inc %edi // Skip flags+len byte.
1386 andb $0x1f,%al // Just the length, not the flags.
1387 add %eax,%edi // Skip the name.
1388 addl $3,%edi // The codeword is 4-byte aligned.
1393 COMPILING ----------------------------------------------------------------------
1395 Now we'll talk about how FORTH compiles words. Recall that a word definition looks like this:
1399 and we have to turn this into:
1401 pointer to previous word
1404 +--|------+---+---+---+---+---+---+---+---+------------+------------+------------+------------+
1405 | LINK | 6 | D | O | U | B | L | E | 0 | DOCOL | DUP | + | EXIT |
1406 +---------+---+---+---+---+---+---+---+---+------------+--|---------+------------+------------+
1407 ^ len pad codeword |
1409 LATEST points here points to codeword of DUP
1411 There are several problems to solve. Where to put the new word? How do we read words? How
1412 do we define the words : (COLON) and ; (SEMICOLON)?
1414 FORTH solves this rather elegantly and as you might expect in a very low-level way which
1415 allows you to change how the compiler works on your own code.
1417 FORTH has an INTERPRETER function (a true interpreter this time, not DOCOL) which runs in a
1418 loop, reading words (using WORD), looking them up (using FIND), turning them into codeword
1419 pointers (using >CFA) and deciding what to do with them.
1421 What it does depends on the mode of the interpreter (in variable STATE).
1423 When STATE is zero, the interpreter just runs each word as it looks them up. This is known as
1426 The interesting stuff happens when STATE is non-zero -- compiling mode. In this mode the
1427 interpreter appends the codeword pointer to user memory (the HERE variable points to the next
1428 free byte of user memory).
1430 So you may be able to see how we could define : (COLON). The general plan is:
1432 (1) Use WORD to read the name of the function being defined.
1434 (2) Construct the dictionary entry -- just the header part -- in user memory:
1436 pointer to previous word (from LATEST) +-- Afterwards, HERE points here, where
1437 ^ | the interpreter will start appending
1439 +--|------+---+---+---+---+---+---+---+---+------------+
1440 | LINK | 6 | D | O | U | B | L | E | 0 | DOCOL |
1441 +---------+---+---+---+---+---+---+---+---+------------+
1444 (3) Set LATEST to point to the newly defined word, ...
1446 (4) .. and most importantly leave HERE pointing just after the new codeword. This is where
1447 the interpreter will append codewords.
1449 (5) Set STATE to 1. This goes into compile mode so the interpreter starts appending codewords to
1450 our partially-formed header.
1452 After : has run, our input is here:
1457 Next byte returned by KEY
1459 so the interpreter (now it's in compile mode, so I guess it's really the compiler) reads DUP,
1460 gets its codeword pointer, and appends it:
1462 +-- HERE updated to point here.
1465 +---------+---+---+---+---+---+---+---+---+------------+------------+
1466 | LINK | 6 | D | O | U | B | L | E | 0 | DOCOL | DUP |
1467 +---------+---+---+---+---+---+---+---+---+------------+------------+
1470 Next we read +, get the codeword pointer, and append it:
1472 +-- HERE updated to point here.
1475 +---------+---+---+---+---+---+---+---+---+------------+------------+------------+
1476 | LINK | 6 | D | O | U | B | L | E | 0 | DOCOL | DUP | + |
1477 +---------+---+---+---+---+---+---+---+---+------------+------------+------------+
1480 The issue is what happens next. Obviously what we _don't_ want to happen is that we
1481 read ";" and compile it and go on compiling everything afterwards.
1483 At this point, FORTH uses a trick. Remember the length byte in the dictionary definition
1484 isn't just a plain length byte, but can also contain flags. One flag is called the
1485 IMMEDIATE flag (F_IMMED in this code). If a word in the dictionary is flagged as
1486 IMMEDIATE then the interpreter runs it immediately _even if it's in compile mode_.
1488 I hope I don't need to explain that ; (SEMICOLON) is just such a word, flagged as IMMEDIATE.
1489 And all it does is append the codeword for EXIT on to the current definition and switch
1490 back to immediate mode (set STATE back to 0). Shortly we'll see the actual definition
1491 of ; and we'll see that it's really a very simple definition, declared IMMEDIATE.
1493 After the interpreter reads ; and executes it 'immediately', we get this:
1495 +---------+---+---+---+---+---+---+---+---+------------+------------+------------+------------+
1496 | LINK | 6 | D | O | U | B | L | E | 0 | DOCOL | DUP | + | EXIT |
1497 +---------+---+---+---+---+---+---+---+---+------------+------------+------------+------------+
1504 And that's it, job done, our new definition is compiled, and we're back in immediate mode
1505 just reading and executing words, perhaps including a call to test our new word DOUBLE.
1507 The only last wrinkle in this is that while our word was being compiled, it was in a
1508 half-finished state. We certainly wouldn't want DOUBLE to be called somehow during
1509 this time. There are several ways to stop this from happening, but in FORTH what we
1510 do is flag the word with the HIDDEN flag (F_HIDDEN in this code) just while it is
1511 being compiled. This prevents FIND from finding it, and thus in theory stops any
1512 chance of it being called.
1514 Compared to the description above, the actual definition of : (COLON) is comparatively simple:
1517 defcode ":",1,,COLON
1519 // Get the word and create a dictionary entry header for it.
1520 call _WORD // Returns %ecx = length, %edi = pointer to word.
1521 mov %edi,%ebx // %ebx = address of the word
1523 movl var_HERE,%edi // %edi is the address of the header
1524 movl var_LATEST,%eax // Get link pointer
1525 stosl // and store it in the header.
1527 mov %cl,%al // Get the length.
1528 orb $F_HIDDEN,%al // Set the HIDDEN flag on this entry.
1529 stosb // Store the length/flags byte.
1531 mov %ebx,%esi // %esi = word
1532 rep movsb // Copy the word
1534 addl $3,%edi // Align to next 4 byte boundary.
1537 movl $DOCOL,%eax // The codeword for user-created words is always DOCOL (the interpreter)
1540 // Header built, so now update LATEST and HERE.
1541 // We'll be compiling words and putting them HERE.
1543 movl %eax,var_LATEST
1546 // And go into compile mode by setting STATE to 1.
1551 , (COMMA) is a standard FORTH word which appends a 32 bit integer (normally a codeword
1552 pointer) to the user data area pointed to by HERE, and adds 4 to HERE.
1555 defcode ",",1,,COMMA
1556 pop %eax // Code pointer to store.
1560 movl var_HERE,%edi // HERE
1562 movl %edi,var_HERE // Update HERE (incremented)
1566 ; (SEMICOLON) is also elegantly simple. Notice the F_IMMED flag.
1569 defcode ";",1,F_IMMED,SEMICOLON
1570 movl $EXIT,%eax // EXIT is the final codeword in compiled words.
1571 call _COMMA // Store it.
1572 call _HIDDEN // Toggle the HIDDEN flag (unhides the new word).
1573 xor %eax,%eax // Set STATE to 0 (back to execute mode).
1578 EXTENDING THE COMPILER ----------------------------------------------------------------------
1580 Words flagged with IMMEDIATE (F_IMMED) aren't just for the FORTH compiler to use. You can define
1581 your own IMMEDIATE words too, and this is a crucial aspect when extending basic FORTH, because
1582 it allows you in effect to extend the compiler itself. Does gcc let you do that?
1584 Standard FORTH words like IF, WHILE, .", [ and so on are all written as extensions to the basic
1585 compiler, and are all IMMEDIATE words.
1587 The IMMEDIATE word toggles the F_IMMED (IMMEDIATE flag) on the most recently defined word,
1588 or on the current word if you call it in the middle of a definition.
1592 : MYIMMEDWORD IMMEDIATE
1596 but some FORTH programmers write this instead:
1602 The two usages are equivalent, to a first approximation.
1605 defcode "IMMEDIATE",9,F_IMMED,IMMEDIATE
1609 movl var_LATEST,%edi // LATEST word.
1610 addl $4,%edi // Point to name/flags byte.
1611 xorb $F_IMMED,(%edi) // Toggle the IMMED bit.
1615 HIDDEN toggles the other flag, F_HIDDEN, of the latest word. Note that words flagged
1616 as hidden are defined but cannot be called, so this is rarely used.
1619 defcode "HIDDEN",6,,HIDDEN
1623 movl var_LATEST,%edi // LATEST word.
1624 addl $4,%edi // Point to name/flags byte.
1625 xorb $F_HIDDEN,(%edi) // Toggle the HIDDEN bit.
1629 ' (TICK) is a standard FORTH word which returns the codeword pointer of the next word.
1631 The common usage is:
1635 which appends the codeword of FOO to the current word we are defining (this only works in compiled code).
1637 You tend to use ' in IMMEDIATE words. For example an alternate (and rather useless) way to define
1638 a literal 2 might be:
1641 ' LIT , \ Appends LIT to the currently-being-defined word
1642 2 , \ Appends the number 2 to the currently-being-defined word
1649 (If you don't understand how LIT2 works, then you should review the material about compiling words
1650 and immediate mode).
1652 This definition of ' uses a cheat which I copied from buzzard92. As a result it only works in
1653 compiled code. It is possible to write a version of ' based on WORD, FIND, >CFA which works in
1657 lodsl // Get the address of the next word and skip it.
1658 pushl %eax // Push it on the stack.
1662 BRANCHING ----------------------------------------------------------------------
1664 It turns out that all you need in order to define looping constructs, IF-statements, etc.
1667 BRANCH is an unconditional branch. 0BRANCH is a conditional branch (it only branches if the
1668 top of stack is zero).
1670 The diagram below shows how BRANCH works in some imaginary compiled word. When BRANCH executes,
1671 %esi starts by pointing to the offset field (compare to LIT above):
1673 +---------------------+-------+---- - - ---+------------+------------+---- - - - ----+------------+
1674 | (Dictionary header) | DOCOL | | BRANCH | offset | (skipped) | word |
1675 +---------------------+-------+---- - - ---+------------+-----|------+---- - - - ----+------------+
1678 | +-----------------------+
1679 %esi added to offset
1681 The offset is added to %esi to make the new %esi, and the result is that when NEXT runs, execution
1682 continues at the branch target. Negative offsets work as expected.
1684 0BRANCH is the same except the branch happens conditionally.
1686 Now standard FORTH words such as IF, THEN, ELSE, WHILE, REPEAT, etc. can be implemented entirely
1687 in FORTH. They are IMMEDIATE words which append various combinations of BRANCH or 0BRANCH
1688 into the word currently being compiled.
1690 As an example, code written like this:
1692 condition-code IF true-part THEN rest-code
1696 condition-code 0BRANCH OFFSET true-part rest-code
1702 defcode "BRANCH",6,,BRANCH
1703 add (%esi),%esi // add the offset to the instruction pointer
1706 defcode "0BRANCH",7,,ZBRANCH
1708 test %eax,%eax // top of stack is zero?
1709 jz code_BRANCH // if so, jump back to the branch function above
1710 lodsl // otherwise we need to skip the offset
1714 PRINTING STRINGS ----------------------------------------------------------------------
1716 LITSTRING and EMITSTRING are primitives used to implement the ." operator (which is
1717 written in FORTH). See the definition of that operator below.
1720 defcode "LITSTRING",9,,LITSTRING
1721 lodsl // get the length of the string
1722 push %eax // push it on the stack
1723 push %esi // push the address of the start of the string
1724 addl %eax,%esi // skip past the string
1725 addl $3,%esi // but round up to next 4 byte boundary
1729 defcode "EMITSTRING",10,,EMITSTRING
1730 mov $1,%ebx // 1st param: stdout
1731 pop %ecx // 2nd param: address of string
1732 pop %edx // 3rd param: length of string
1733 mov $__NR_write,%eax // write syscall
1738 COLD START AND INTERPRETER ----------------------------------------------------------------------
1740 COLD is the first FORTH function called, almost immediately after the FORTH system "boots".
1742 INTERPRETER is the FORTH interpreter ("toploop", "toplevel" or "REPL" might be a more accurate
1743 description -- see: http://en.wikipedia.org/wiki/REPL).
1747 // COLD must not return (ie. must not call EXIT).
1748 defword "COLD",4,,COLD
1749 .int INTERPRETER // call the interpreter loop (never returns)
1750 .int LIT,1,SYSEXIT // hmmm, but in case it does, exit(1).
1752 /* This interpreter is pretty simple, but remember that in FORTH you can always override
1753 * it later with a more powerful one!
1755 defword "INTERPRETER",11,,INTERPRETER
1756 .int INTERPRET,RDROP,INTERPRETER
1758 defcode "INTERPRET",9,,INTERPRET
1759 call _WORD // Returns %ecx = length, %edi = pointer to word.
1761 // Is it in the dictionary?
1763 movl %eax,interpret_is_lit // Not a literal number (not yet anyway ...)
1764 call _FIND // Returns %eax = pointer to header or 0 if not found.
1765 test %eax,%eax // Found?
1768 // In the dictionary. Is it an IMMEDIATE codeword?
1769 mov %eax,%edi // %edi = dictionary entry
1770 movb 4(%edi),%al // Get name+flags.
1771 push %ax // Just save it for now.
1772 call _TCFA // Convert dictionary entry (in %edi) to codeword pointer.
1774 andb $F_IMMED,%al // Is IMMED flag set?
1776 jnz 4f // If IMMED, jump straight to executing.
1780 1: // Not in the dictionary (not a word) so assume it's a literal number.
1781 incl interpret_is_lit
1782 call _SNUMBER // Returns the parsed number in %eax
1784 mov $LIT,%eax // The word is LIT
1786 2: // Are we compiling or executing?
1789 jz 4f // Jump if executing.
1791 // Compiling - just append the word to the current dictionary definition.
1793 mov interpret_is_lit,%ecx // Was it a literal?
1796 mov %ebx,%eax // Yes, so LIT is followed by a number.
1800 4: // Executing - run it!
1801 mov interpret_is_lit,%ecx // Literal?
1802 test %ecx,%ecx // Literal?
1805 // Not a literal, execute it now. This never returns, but the codeword will
1806 // eventually call NEXT which will reenter the loop in INTERPRETER.
1809 5: // Executing a literal, which means push it on the stack.
1816 .int 0 // Flag used to record if reading a literal
1819 ODDS AND ENDS ----------------------------------------------------------------------
1821 CHAR puts the ASCII code of the first character of the following word on the stack. For example
1822 CHAR A puts 65 on the stack.
1824 SYSEXIT exits the process using Linux exit syscall.
1827 defcode "CHAR",4,,CHAR
1828 call _WORD // Returns %ecx = length, %edi = pointer to word.
1830 movb (%edi),%al // Get the first character of the word.
1831 push %eax // Push it onto the stack.
1834 // NB: SYSEXIT must be the last entry in the built-in dictionary.
1835 defcode SYSEXIT,7,,SYSEXIT
1841 START OF FORTH CODE ----------------------------------------------------------------------
1843 We've now reached the stage where the FORTH system is running and self-hosting. All further
1844 words can be written as FORTH itself, including words like IF, THEN, .", etc which in most
1845 languages would be considered rather fundamental.
1847 As a kind of trick, I prefill the input buffer with the initial FORTH code. Once this code
1848 has run (when we get to the "OK" prompt), this input buffer is reused for reading any further
1851 Some notes about the code:
1853 \ (backslash) is the FORTH way to start a comment which goes up to the next newline. However
1854 because this is a C-style string, I have to escape the backslash, which is why they appear as
1857 Similarly, any backslashes in the code are doubled, and " becomes \" (eg. the definition of ."
1858 is written as : .\" ... ;)
1860 I use indenting to show structure. The amount of whitespace has no meaning to FORTH however
1861 except that you must use at least one whitespace character between words, and words themselves
1862 cannot contain whitespace.
1864 FORTH is case-sensitive. Use capslock!
1872 // Multi-line constant gives 'Warning: unterminated string; newline inserted' messages which you can ignore.
1874 \\ Define some character constants
1880 \\ CR prints a carriage return
1883 \\ SPACE prints a space
1884 : SPACE 'SPACE' EMIT ;
1886 \\ Primitive . (DOT) function doesn't follow with a blank, so redefine it to behave like FORTH.
1887 \\ Notice how we can trivially redefine existing functions.
1890 \\ DUP, DROP are defined in assembly for speed, but this is how you might define them
1891 \\ in FORTH. Notice use of the scratch variables _X and _Y.
1892 \\ : DUP _X ! _X @ _X @ ;
1895 \\ The 2... versions of the standard operators work on pairs of stack entries. They're not used
1896 \\ very commonly so not really worth writing in assembler. Here is how they are defined in FORTH.
1900 \\ More standard FORTH words.
1904 \\ [ and ] allow you to break into immediate mode while compiling a word.
1905 : [ IMMEDIATE \\ define [ as an immediate word
1906 0 STATE ! \\ go into immediate mode
1910 1 STATE ! \\ go back to compile mode
1913 \\ LITERAL takes whatever is on the stack and compiles LIT <foo>
1915 ' LIT , \\ compile LIT
1916 , \\ compile the literal itself (from the stack)
1919 \\ condition IF true-part THEN rest
1921 \\ condition 0BRANCH OFFSET true-part rest
1922 \\ where OFFSET is the offset of 'rest'
1923 \\ condition IF true-part ELSE false-part THEN
1925 \\ condition 0BRANCH OFFSET true-part BRANCH OFFSET2 false-part rest
1926 \\ where OFFSET if the offset of false-part and OFFSET2 is the offset of rest
1928 \\ IF is an IMMEDIATE word which compiles 0BRANCH followed by a dummy offset, and places
1929 \\ the address of the 0BRANCH on the stack. Later when we see THEN, we pop that address
1930 \\ off the stack, calculate the offset, and back-fill the offset.
1932 ' 0BRANCH , \\ compile 0BRANCH
1933 HERE @ \\ save location of the offset on the stack
1934 0 , \\ compile a dummy offset
1939 HERE @ SWAP - \\ calculate the offset from the address saved on the stack
1940 SWAP ! \\ store the offset in the back-filled location
1944 ' BRANCH , \\ definite branch to just over the false-part
1945 HERE @ \\ save location of the offset on the stack
1946 0 , \\ compile a dummy offset
1947 SWAP \\ now back-fill the original (IF) offset
1948 DUP \\ same as for THEN word above
1953 \\ BEGIN loop-part condition UNTIL
1955 \\ loop-part condition 0BRANCH OFFSET
1956 \\ where OFFSET points back to the loop-part
1957 \\ This is like do { loop-part } while (condition) in the C language
1959 HERE @ \\ save location on the stack
1963 ' 0BRANCH , \\ compile 0BRANCH
1964 HERE @ - \\ calculate the offset from the address saved on the stack
1965 , \\ compile the offset here
1968 \\ BEGIN loop-part AGAIN
1970 \\ loop-part BRANCH OFFSET
1971 \\ where OFFSET points back to the loop-part
1972 \\ In other words, an infinite loop which can only be returned from with EXIT
1974 ' BRANCH , \\ compile BRANCH
1975 HERE @ - \\ calculate the offset back
1976 , \\ compile the offset here
1979 \\ BEGIN condition WHILE loop-part REPEAT
1981 \\ condition 0BRANCH OFFSET2 loop-part BRANCH OFFSET
1982 \\ where OFFSET points back to condition (the beginning) and OFFSET2 points to after the whole piece of code
1983 \\ So this is like a while (condition) { loop-part } loop in the C language
1985 ' 0BRANCH , \\ compile 0BRANCH
1986 HERE @ \\ save location of the offset2 on the stack
1987 0 , \\ compile a dummy offset2
1991 ' BRANCH , \\ compile BRANCH
1992 SWAP \\ get the original offset (from BEGIN)
1993 HERE @ - , \\ and compile it after BRANCH
1995 HERE @ SWAP - \\ calculate the offset2
1996 SWAP ! \\ and back-fill it in the original location
1999 \\ With the looping constructs, we can now write SPACES, which writes n spaces to stdout.
2002 SPACE \\ print a space
2003 1- \\ until we count down to 0
2008 \\ .S prints the contents of the stack. Very useful for debugging.
2010 DSP@ \\ get current stack pointer
2012 DUP @ . \\ print the stack element
2014 DUP S0 @ 4- = \\ stop when we get to the top
2019 \\ DEPTH returns the depth of the stack.
2020 : DEPTH S0 @ DSP@ - ;
2022 \\ .\" is the print string operator in FORTH. Example: .\" Something to print\"
2023 \\ The space after the operator is the ordinary space required between words.
2024 \\ This is tricky to define because it has to do different things depending on whether
2025 \\ we are compiling or in immediate mode. (Thus the word is marked IMMEDIATE so it can
2026 \\ detect this and do different things).
2027 \\ In immediate mode we just keep reading characters and printing them until we get to
2028 \\ the next double quote.
2029 \\ In compile mode we have the problem of where we're going to store the string (remember
2030 \\ that the input buffer where the string comes from may be overwritten by the time we
2031 \\ come round to running the function). We store the string in the compiled function
2033 \\ ..., LITSTRING, string length, string rounded up to 4 bytes, EMITSTRING, ...
2035 STATE @ \\ compiling?
2037 ' LITSTRING , \\ compile LITSTRING
2038 HERE @ \\ save the address of the length word on the stack
2039 0 , \\ dummy length - we don't know what it is yet
2041 KEY \\ get next character of the string
2044 HERE @ !b \\ store the character in the compiled image
2045 1 HERE +! \\ increment HERE pointer by 1 byte
2047 DROP \\ drop the double quote character at the end
2048 DUP \\ get the saved address of the length word
2049 HERE @ SWAP - \\ calculate the length
2050 4- \\ subtract 4 (because we measured from the start of the length word)
2051 SWAP ! \\ and back-fill the length location
2052 HERE @ \\ round up to next multiple of 4 bytes for the remaining code
2056 ' EMITSTRING , \\ compile the final EMITSTRING
2058 \\ In immediate mode, just read characters and print them until we get
2059 \\ to the ending double quote. Much simpler than the above code!
2062 DUP '\"' = IF EXIT THEN
2068 \\ While compiling, [COMPILE] WORD compiles WORD if it would otherwise be IMMEDIATE.
2069 : [COMPILE] IMMEDIATE
2070 WORD \\ get the next word
2071 FIND \\ find it in the dictionary
2072 >CFA \\ get its codeword
2073 , \\ and compile that
2076 \\ RECURSE makes a recursive call to the current word that is being compiled.
2077 \\ Normally while a word is being compiled, it is marked HIDDEN so that references to the
2078 \\ same word within are calls to the previous definition of the word.
2080 LATEST @ >CFA \\ LATEST points to the word being compiled at the moment
2084 \\ ALLOT is used to allocate (static) memory when compiling. It increases HERE by
2085 \\ the amount given on the stack.
2089 \\ Finally print the welcome prompt.
2090 .\" JONESFORTH VERSION \" VERSION @ . CR
2103 /* END OF jonesforth.S */
git.annexia.org / jonesforth.git / blob