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
4 gcc -m32 -nostdlib -static -Wl,-Ttext,0 -o jonesforth jonesforth.S
6 INTRODUCTION ----------------------------------------------------------------------
8 FORTH is one of those alien languages which most working programmers regard in the same
9 way as Haskell, LISP, and so on. Something so strange that they'd rather any thoughts
10 of it just go away so they can get on with writing this paying code. But that's wrong
11 and if you care at all about programming then you should at least understand all these
12 languages, even if you will never use them.
14 LISP is the ultimate high-level language, and features from LISP are being added every
15 decade to the more common languages. But FORTH is in some ways the ultimate in low level
16 programming. Out of the box it lacks features like dynamic memory management and even
17 strings. In fact, at its primitive level it lacks even basic concepts like IF-statements
20 Why then would you want to learn FORTH? There are several very good reasons. First
21 and foremost, FORTH is minimal. You really can write a complete FORTH in, say, 2000
22 lines of code. I don't just mean a FORTH program, I mean a complete FORTH operating
23 system, environment and language. You could boot such a FORTH on a bare PC and it would
24 come up with a prompt where you could start doing useful work. The FORTH you have here
25 isn't minimal and uses a Linux process as its 'base PC' (both for the purposes of making
26 it a good tutorial). It's possible to completely understand the system. Who can say they
27 completely understand how Linux works, or gcc?
29 Secondly FORTH has a peculiar bootstrapping property. By that I mean that after writing
30 a little bit of assembly to talk to the hardware and implement a few primitives, all the
31 rest of the language and compiler is written in FORTH itself. Remember I said before
32 that FORTH lacked IF-statements and loops? Well of course it doesn't really because
33 such a lanuage would be useless, but my point was rather that IF-statements and loops are
34 written in FORTH itself.
36 Now of course this is common in other languages as well, and in those languages we call
37 them 'libraries'. For example in C, 'printf' is a library function written in C. But
38 in FORTH this goes way beyond mere libraries. Can you imagine writing C's 'if' in C?
39 And that brings me to my third reason: If you can write 'if' in FORTH, then why restrict
40 yourself to the usual if/while/for/switch constructs? You want a construct that iterates
41 over every other element in a list of numbers? You can add it to the language. What
42 about an operator which pulls in variables directly from a configuration file and makes
43 them available as FORTH variables? Or how about adding Makefile-like dependencies to
44 the language? No problem in FORTH. This concept isn't common in programming languages,
45 but it has a name (in fact two names): "macros" (by which I mean LISP-style macros, not
46 the lame C preprocessor) and "domain specific languages" (DSLs).
48 This tutorial isn't about learning FORTH as the language. I'll point you to some references
49 you should read if you're not familiar with using FORTH. This tutorial is about how to
50 write FORTH. In fact, until you understand how FORTH is written, you'll have only a very
51 superficial understanding of how to use it.
53 So if you're not familiar with FORTH or want to refresh your memory here are some online
56 http://en.wikipedia.org/wiki/Forth_%28programming_language%29
58 http://galileo.phys.virginia.edu/classes/551.jvn.fall01/primer.htm
60 http://wiki.laptop.org/go/Forth_Lessons
62 Here is another "Why FORTH?" essay: http://www.jwdt.com/~paysan/why-forth.html
64 ACKNOWLEDGEMENTS ----------------------------------------------------------------------
66 This code draws heavily on the design of LINA FORTH (http://home.hccnet.nl/a.w.m.van.der.horst/lina.html)
67 by Albert van der Horst. Any similarities in the code are probably not accidental.
69 Also I used this document (http://ftp.funet.fi/pub/doc/IOCCC/1992/buzzard.2.design) which really
70 defies easy explanation.
72 SETTING UP ----------------------------------------------------------------------
74 Let's get a few housekeeping things out of the way. Firstly because I need to draw lots of
75 ASCII-art diagrams to explain concepts, the best way to look at this is using a window which
76 uses a fixed width font and is at least this wide:
78 <------------------------------------------------------------------------------------------------------------------------>
80 Secondly make sure TABS are set to 8 characters. The following should be a vertical
81 line. If not, sort out your tabs.
87 Thirdly I assume that your screen is at least 50 characters high.
89 ASSEMBLING ----------------------------------------------------------------------
91 If you want to actually run this FORTH, rather than just read it, you will need Linux on an
92 i386. Linux because instead of programming directly to the hardware on a bare PC which I
93 could have done, I went for a simpler tutorial by assuming that the 'hardware' is a Linux
94 process with a few basic system calls (read, write and exit and that's about all). i386
95 is needed because I had to write the assembly for a processor, and i386 is by far the most
96 common. (Of course when I say 'i386', any 32- or 64-bit x86 processor will do. I'm compiling
97 this on a 64 bit AMD Opteron).
99 Again, to assemble this you will need gcc and gas (the GNU assembler). The commands to
100 assemble and run the code (save this file as 'jonesforth.S') are:
102 gcc -m32 -nostdlib -static -Wl,-Ttext,0 -o jonesforth jonesforth.S
105 You will see lots of 'Warning: unterminated string; newline inserted' messages from the
106 assembler. That's just because the GNU assembler doesn't have a good syntax for multi-line
107 strings (or rather it used to, but the developers removed it!) so I've abused the syntax
108 slightly to make things readable. Ignore these warnings.
110 If you want to run your own FORTH programs you can do:
112 ./jonesforth < myprog.f
114 If you want to load your own FORTH code and then continue reading user commands, you can do:
116 cat myfunctions.f - | ./jonesforth
118 ASSEMBLER ----------------------------------------------------------------------
120 (You can just skip to the next section -- you don't need to be able to read assembler to
121 follow this tutorial).
123 However if you do want to read the assembly code here are a few notes about gas (the GNU assembler):
125 (1) Register names are prefixed with '%', so %eax is the 32 bit i386 accumulator. The registers
126 available on i386 are: %eax, %ebx, %ecx, %edx, %esi, %edi, %ebp and %esp, and most of them
127 have special purposes.
129 (2) Add, mov, etc. take arguments in the form SRC,DEST. So mov %eax,%ecx moves %eax -> %ecx
131 (3) Constants are prefixed with '$', and you mustn't forget it! If you forget it then it
132 causes a read from memory instead, so:
133 mov $2,%eax moves number 2 into %eax
134 mov 2,%eax reads the 32 bit word from address 2 into %eax (ie. most likely a mistake)
136 (4) gas has a funky syntax for local labels, where '1f' (etc.) means label '1:' "forwards"
137 and '1b' (etc.) means label '1:' "backwards".
139 (5) 'ja' is "jump if above", 'jb' for "jump if below", 'je' "jump if equal" etc.
141 (6) gas has a reasonably nice .macro syntax, and I use them a lot to make the code shorter and
144 For more help reading the assembler, do "info gas" at the Linux prompt.
146 Now the tutorial starts in earnest.
148 THE DICTIONARY ----------------------------------------------------------------------
150 In FORTH as you will know, functions are called "words", as just as in other languages they
151 have a name and a definition. Here are two FORTH words:
153 : DOUBLE DUP + ; \ name is "DOUBLE", definition is "DUP +"
154 : QUADRUPLE DOUBLE DOUBLE ; \ name is "QUADRUPLE", definition is "DOUBLE DOUBLE"
156 Words, both built-in ones and ones which the programmer defines later, are stored in a dictionary
157 which is just a linked list of dictionary entries.
159 <--- DICTIONARY ENTRY (HEADER) ----------------------->
160 +------------------------+--------+---------- - - - - +----------- - - - -
161 | LINK POINTER | LENGTH/| NAME | DEFINITION
163 +--- (4 bytes) ----------+- byte -+- n bytes - - - - +----------- - - - -
165 I'll come to the definition of the word later. For now just look at the header. The first
166 4 bytes are the link pointer. This points back to the previous word in the dictionary, or, for
167 the first word in the dictionary it is just a NULL pointer. Then comes a length/flags byte.
168 The length of the word can be up to 31 characters (5 bits used) and the top three bits are used
169 for various flags which I'll come to later. This is followed by the name itself, and in this
170 implementation the name is rounded up to a multiple of 4 bytes by padding it with zero bytes.
171 That's just to ensure that the definition starts on a 32 bit boundary.
173 A FORTH variable called LATEST contains a pointer to the most recently defined word, in
174 other words, the head of this linked list.
176 DOUBLE and QUADRUPLE might look like this:
178 pointer to previous word
181 +--|------+---+---+---+---+---+---+---+---+------------- - - - -
182 | LINK | 6 | D | O | U | B | L | E | 0 | (definition ...)
183 +---------+---+---+---+---+---+---+---+---+------------- - - - -
186 +--|------+---+---+---+---+---+---+---+---+---+---+---+---+------------- - - - -
187 | LINK | 9 | Q | U | A | D | R | U | P | L | E | 0 | 0 | (definition ...)
188 +---------+---+---+---+---+---+---+---+---+---+---+---+---+------------- - - - -
194 You shoud be able to see from this how you might implement functions to find a word in
195 the dictionary (just walk along the dictionary entries starting at LATEST and matching
196 the names until you either find a match or hit the NULL pointer at the end of the dictionary),
197 and add a word to the dictionary (create a new definition, set its LINK to LATEST, and set
198 LATEST to point to the new word). We'll see precisely these functions implemented in
199 assembly code later on.
201 One interesting consequence of using a linked list is that you can redefine words, and
202 a newer definition of a word overrides an older one. This is an important concept in
203 FORTH because it means that any word (even "built-in" or "standard" words) can be
204 overridden with a new definition, either to enhance it, to make it faster or even to
205 disable it. However because of the way that FORTH words get compiled, which you'll
206 understand below, words defined using the old definition of a word continue to use
207 the old definition. Only words defined after the new definition use the new definition.
209 DIRECT THREADED CODE ----------------------------------------------------------------------
211 Now we'll get to the really crucial bit in understanding FORTH, so go and get a cup of tea
212 or coffee and settle down. It's fair to say that if you don't understand this section, then you
213 won't "get" how FORTH works, and that would be a failure on my part for not explaining it well.
214 So if after reading this section a few times you don't understand it, please email me
217 Let's talk first about what "threaded code" means. Imagine a peculiar version of C where
218 you are only allowed to call functions without arguments. (Don't worry for now that such a
219 language would be completely useless!) So in our peculiar C, code would look like this:
228 and so on. How would a function, say 'f' above, be compiled by a standard C compiler?
229 Probably into assembly code like this. On the right hand side I've written the actual
233 CALL a E8 08 00 00 00
234 CALL b E8 1C 00 00 00
235 CALL c E8 2C 00 00 00
236 ; ignore the return from the function for now
238 "E8" is the x86 machine code to "CALL" a function. In the first 20 years of computing
239 memory was hideously expensive and we might have worried about the wasted space being used
240 by the repeated "E8" bytes. We can save 20% in code size (and therefore, in expensive memory)
241 by compressing this into just:
243 08 00 00 00 Just the function addresses, without
244 1C 00 00 00 the CALL prefix.
247 [Historical note: If the execution model that FORTH uses looks strange from the following
248 paragraphs, then it was motivated entirely by the need to save memory on early computers.
249 This code compression isn't so important now when our machines have more memory in their L1
250 caches than those early computers had in total, but the execution model still has some
253 Of course this code won't run directly any more. Instead we need to write an interpreter
254 which takes each pair of bytes and calls it.
256 On an i386 machine it turns out that we can write this interpreter rather easily, in just
257 two assembly instructions which turn into just 3 bytes of machine code. Let's store the
258 pointer to the next word to execute in the %esi register:
260 08 00 00 00 <- We're executing this one now. %esi is the _next_ one to execute.
264 The all-important x86 instruction is called LODSL (or in Intel manuals, LODSW). It does
265 two things. Firstly it reads the memory at %esi into the accumulator (%eax). Secondly it
266 increments %esi by 4 bytes. So after LODSL, the situation now looks like this:
268 08 00 00 00 <- We're still executing this one
269 1C 00 00 00 <- %eax now contains this address (0x0000001C)
272 Now we just need to jump to the address in %eax. This is again just a single x86 instruction
273 written JMP *(%eax). And after doing the jump, the situation looks like:
276 1C 00 00 00 <- Now we're executing this subroutine.
279 To make this work, each subroutine is followed by the two instructions 'LODSL; JMP *(%eax)'
280 which literally make the jump to the next subroutine.
282 And that brings us to our first piece of actual code! Well, it's a macro.
291 /* The macro is called NEXT. That's a FORTH-ism. It expands to those two instructions.
293 Every FORTH primitive that we write has to be ended by NEXT. Think of it kind of like
296 The above describes what is known as direct threaded code.
298 To sum up: We compress our function calls down to a list of addresses and use a somewhat
299 magical macro to act as a "jump to next function in the list". We also use one register (%esi)
300 to act as a kind of instruction pointer, pointing to the next function in the list.
302 I'll just give you a hint of what is to come by saying that a FORTH definition such as:
304 : QUADRUPLE DOUBLE DOUBLE ;
306 actually compiles (almost, not precisely but we'll see why in a moment) to a list of
307 function addresses for DOUBLE, DOUBLE and a special function called EXIT to finish off.
309 At this point, REALLY EAGLE-EYED ASSEMBLY EXPERTS are saying "JONES, YOU'VE MADE A MISTAKE!".
311 I lied about JMP *(%eax).
313 INDIRECT THREADED CODE ----------------------------------------------------------------------
315 It turns out that direct threaded code is interesting but only if you want to just execute
316 a list of functions written in assembly language. So QUADRUPLE would work only if DOUBLE
317 was an assembly language function. In the direct threaded code, QUADRUPLE would look like:
320 | addr of DOUBLE --------------------> (assembly code to do the double)
321 +------------------+ NEXT
322 %esi -> | addr of DOUBLE |
325 We can add an extra indirection to allow us to run both words written in assembly language
326 (primitives written for speed) and words written in FORTH themselves as lists of addresses.
328 The extra indirection is the reason for the brackets in JMP *(%eax).
330 Let's have a look at how QUADRUPLE and DOUBLE really look in FORTH:
332 : QUADRUPLE DOUBLE DOUBLE ;
335 | codeword | : DOUBLE DUP + ;
337 | addr of DOUBLE ---------------> +------------------+
338 +------------------+ | codeword |
339 | addr of DOUBLE | +------------------+
340 +------------------+ | addr of DUP --------------> +------------------+
341 | addr of EXIT | +------------------+ | codeword -------+
342 +------------------+ %esi -> | addr of + --------+ +------------------+ |
343 +------------------+ | | assembly to <-----+
344 | addr of EXIT | | | implement DUP |
345 +------------------+ | | .. |
348 | +------------------+
350 +-----> +------------------+
352 +------------------+ |
353 | assembly to <------+
360 This is the part where you may need an extra cup of tea/coffee/favourite caffeinated
361 beverage. What has changed is that I've added an extra pointer to the beginning of
362 the definitions. In FORTH this is sometimes called the "codeword". The codeword is
363 a pointer to the interpreter to run the function. For primitives written in
364 assembly language, the "interpreter" just points to the actual assembly code itself.
365 They don't need interpreting, they just run.
367 In words written in FORTH (like QUADRUPLE and DOUBLE), the codeword points to an interpreter
370 I'll show you the interpreter function shortly, but let's recall our indirect
371 JMP *(%eax) with the "extra" brackets. Take the case where we're executing DOUBLE
372 as shown, and DUP has been called. Note that %esi is pointing to the address of +
374 The assembly code for DUP eventually does a NEXT. That:
376 (1) reads the address of + into %eax %eax points to the codeword of +
377 (2) increments %esi by 4
378 (3) jumps to the indirect %eax jumps to the address in the codeword of +,
379 ie. the assembly code to implement +
384 | addr of DOUBLE ---------------> +------------------+
385 +------------------+ | codeword |
386 | addr of DOUBLE | +------------------+
387 +------------------+ | addr of DUP --------------> +------------------+
388 | addr of EXIT | +------------------+ | codeword -------+
389 +------------------+ | addr of + --------+ +------------------+ |
390 +------------------+ | | assembly to <-----+
391 %esi -> | addr of EXIT | | | implement DUP |
392 +------------------+ | | .. |
395 | +------------------+
397 +-----> +------------------+
399 +------------------+ |
400 now we're | assembly to <------+
401 executing | implement + |
407 So I hope that I've convinced you that NEXT does roughly what you'd expect. This is
408 indirect threaded code.
410 I've glossed over four things. I wonder if you can guess without reading on what they are?
416 My list of four things are: (1) What does "EXIT" do? (2) which is related to (1) is how do
417 you call into a function, ie. how does %esi start off pointing at part of QUADRUPLE, but
418 then point at part of DOUBLE. (3) What goes in the codeword for the words which are written
419 in FORTH? (4) How do you compile a function which does anything except call other functions
420 ie. a function which contains a number like : DOUBLE 2 * ; ?
422 THE INTERPRETER AND RETURN STACK ------------------------------------------------------------
424 Going at these in no particular order, let's talk about issues (3) and (2), the interpreter
425 and the return stack.
427 Words which are defined in FORTH need a codeword which points to a little bit of code to
428 give them a "helping hand" in life. They don't need much, but they do need what is known
429 as an "interpreter", although it doesn't really "interpret" in the same way that, say,
430 Java bytecode used to be interpreted (ie. slowly). This interpreter just sets up a few
431 machine registers so that the word can then execute at full speed using the indirect
432 threaded model above.
434 One of the things that needs to happen when QUADRUPLE calls DOUBLE is that we save the old
435 %esi ("instruction pointer") and create a new one pointing to the first word in DOUBLE.
436 Because we will need to restore the old %esi at the end of DOUBLE (this is, after all, like
437 a function call), we will need a stack to store these "return addresses" (old values of %esi).
439 As you will have read, when reading the background documentation, FORTH has two stacks,
440 an ordinary stack for parameters, and a return stack which is a bit more mysterious. But
441 our return stack is just the stack I talked about in the previous paragraph, used to save
442 %esi when calling from a FORTH word into another FORTH word.
444 In this FORTH, we are using the normal stack pointer (%esp) for the parameter stack.
445 We will use the i386's "other" stack pointer (%ebp, usually called the "frame pointer")
446 for our return stack.
448 I've got two macros which just wrap up the details of using %ebp for the return stack.
449 You use them as for example "PUSHRSP %eax" (push %eax on the return stack) or "POPRSP %ebx"
450 (pop top of return stack into %ebx).
453 /* Macros to deal with the return stack. */
455 lea -4(%ebp),%ebp // push reg on to return stack
460 mov (%ebp),\reg // pop top of return stack to reg
465 And with that we can now talk about the interpreter.
467 In FORTH the interpreter function is often called DOCOL (I think it means "DO COLON" because
468 all FORTH definitions start with a colon, as in : DOUBLE DUP + ;
470 The "interpreter" (it's not really "interpreting") just needs to push the old %esi on the
471 stack and set %esi to the first word in the definition. Remember that we jumped to the
472 function using JMP *(%eax)? Well a consequence of that is that conveniently %eax contains
473 the address of this codeword, so just by adding 4 to it we get the address of the first
474 data word. Finally after setting up %esi, it just does NEXT which causes that first word
478 /* DOCOL - the interpreter! */
482 PUSHRSP %esi // push %esi on to the return stack
483 addl $4,%eax // %eax points to codeword, so make
484 movl %eax,%esi // %esi point to first data word
488 Just to make this absolutely clear, let's see how DOCOL works when jumping from QUADRUPLE
494 +------------------+ DOUBLE:
495 | addr of DOUBLE ---------------> +------------------+
496 +------------------+ %eax -> | addr of DOCOL |
497 %esi -> | addr of DOUBLE | +------------------+
498 +------------------+ | addr of DUP -------------->
499 | addr of EXIT | +------------------+
500 +------------------+ | etc. |
502 First, the call to DOUBLE causes DOCOL (the codeword of DOUBLE). DOCOL does this: It
503 pushes the old %esi on the return stack. %eax points to the codeword of DOUBLE, so we
504 just add 4 on to it to get our new %esi:
509 +------------------+ DOUBLE:
510 | addr of DOUBLE ---------------> +------------------+
511 top of return +------------------+ %eax -> | addr of DOCOL |
512 stack points -> | addr of DOUBLE | + 4 = +------------------+
513 +------------------+ %esi -> | addr of DUP -------------->
514 | addr of EXIT | +------------------+
515 +------------------+ | etc. |
517 Then we do NEXT, and because of the magic of threaded code that increments %esi again
520 Well, it seems to work.
522 One minor point here. Because DOCOL is the first bit of assembly actually to be defined
523 in this file (the others were just macros), and because I usually compile this code with the
524 text segment starting at address 0, DOCOL has address 0. So if you are disassembling the
525 code and see a word with a codeword of 0, you will immediately know that the word is
526 written in FORTH (it's not an assembler primitive) and so uses DOCOL as the interpreter.
528 STARTING UP ----------------------------------------------------------------------
530 Now let's get down to nuts and bolts. When we start the program we need to set up
531 a few things like the return stack. But as soon as we can, we want to jump into FORTH
532 code (albeit much of the "early" FORTH code will still need to be written as
533 assembly language primitives).
535 This is what the set up code does. Does a tiny bit of house-keeping, sets up the
536 separate return stack (NB: Linux gives us the ordinary parameter stack already), then
537 immediately jumps to a FORTH word called COLD. COLD stands for cold-start. In ISO
538 FORTH (but not in this FORTH), COLD can be called at any time to completely reset
539 the state of FORTH, and there is another word called WARM which does a partial reset.
542 /* ELF entry point. */
547 mov %esp,var_S0 // Store the initial data stack pointer.
548 mov $return_stack,%ebp // Initialise the return stack.
550 mov $cold_start,%esi // Initialise interpreter.
551 NEXT // Run interpreter!
554 cold_start: // High-level code without a codeword.
558 We also allocate some space for the return stack and some space to store user
559 definitions. These are static memory allocations using fixed-size buffers, but it
560 wouldn't be a great deal of work to make them dynamic.
564 /* FORTH return stack. */
565 #define RETURN_STACK_SIZE 8192
567 .space RETURN_STACK_SIZE
568 return_stack: // Initial top of return stack.
570 /* Space for user-defined words. */
571 #define USER_DEFS_SIZE 16384
574 .space USER_DEFS_SIZE
577 BUILT-IN WORDS ----------------------------------------------------------------------
579 Remember our dictionary entries (headers). Let's bring those together with the codeword
580 and data words to see how : DOUBLE DUP + ; really looks in memory.
582 pointer to previous word
585 +--|------+---+---+---+---+---+---+---+---+------------+------------+------------+------------+
586 | LINK | 6 | D | O | U | B | L | E | 0 | DOCOL | DUP | + | EXIT |
587 +---------+---+---+---+---+---+---+---+---+------------+--|---------+------------+------------+
590 LINK in next word points to codeword of DUP
592 Initially we can't just write ": DOUBLE DUP + ;" (ie. that literal string) here because we
593 don't yet have anything to read the string, break it up at spaces, parse each word, etc. etc.
594 So instead we will have to define built-in words using the GNU assembler data constructors
595 (like .int, .byte, .string, .ascii and so on -- look them up in the gas info page if you are
598 The long way would be:
599 .int <link to previous word>
601 .ascii "DOUBLE" // string
603 DOUBLE: .int DOCOL // codeword
604 .int DUP // pointer to codeword of DUP
605 .int PLUS // pointer to codeword of +
606 .int EXIT // pointer to codeword of EXIT
608 That's going to get quite tedious rather quickly, so here I define an assembler macro
609 so that I can just write:
611 defword "DOUBLE",6,,DOUBLE
614 and I'll get exactly the same effect.
616 Don't worry too much about the exact implementation details of this macro - it's complicated!
619 /* Flags - these are discussed later. */
621 #define F_HIDDEN 0x20
623 // Store the chain of links.
626 .macro defword name, namelen, flags=0, label
632 .set link,name_\label
633 .byte \flags+\namelen // flags + length byte
634 .ascii "\name" // the name
638 .int DOCOL // codeword - the interpreter
639 // list of word pointers follow
643 Similarly I want a way to write words written in assembly language. There will quite a few
644 of these to start with because, well, everything has to start in assembly before there's
645 enough "infrastructure" to be able to start writing FORTH words, but also I want to define
646 some common FORTH words in assembly language for speed, even though I could write them in FORTH.
648 This is what DUP looks like in memory:
650 pointer to previous word
653 +--|------+---+---+---+---+------------+
654 | LINK | 3 | D | U | P | code_DUP ---------------------> points to the assembly
655 +---------+---+---+---+---+------------+ code used to write DUP,
656 ^ len codeword which is ended with NEXT.
660 Again, for brevity in writing the header I'm going to write an assembler macro called defcode.
663 .macro defcode name, namelen, flags=0, label
669 .set link,name_\label
670 .byte \flags+\namelen // flags + length byte
671 .ascii "\name" // the name
675 .int code_\label // codeword
679 code_\label : // assembler code follows
683 Now some easy FORTH primitives. These are written in assembly for speed. If you understand
684 i386 assembly language then it is worth reading these. However if you don't understand assembly
685 you can skip the details.
689 pop %eax // duplicate top of stack
694 defcode "DROP",4,,DROP
695 pop %eax // drop top of stack
698 defcode "SWAP",4,,SWAP
699 pop %eax // swap top of stack
705 defcode "OVER",4,,OVER
706 mov 4(%esp),%eax // get the second element of stack
707 push %eax // and push it on top
719 defcode "-ROT",4,,NROT
729 incl (%esp) // increment top of stack
733 decl (%esp) // decrement top of stack
736 defcode "4+",2,,INCR4
737 addl $4,(%esp) // increment top of stack
740 defcode "4-",2,,DECR4
741 subl $4,(%esp) // decrement top of stack
745 pop %eax // get top of stack
746 addl %eax,(%esp) // and add it to next word on stack
750 pop %eax // get top of stack
751 subl %eax,(%esp) // and subtract if from next word on stack
758 push %eax // ignore overflow
766 push %eax // push quotient
774 push %edx // push remainder
777 defcode "=",1,,EQU // top two words are equal?
787 defcode "<>",2,,NEQU // top two words are not equal?
797 defcode "0=",2,,ZEQU // top of stack equals 0?
816 defcode "INVERT",6,,INVERT // this is the FORTH "NOT" function
821 RETURNING FROM FORTH WORDS ----------------------------------------------------------------------
823 Time to talk about what happens when we EXIT a function. In this diagram QUADRUPLE has called
824 DOUBLE, and DOUBLE is about to exit (look at where %esi is pointing):
829 +------------------+ DOUBLE
830 | addr of DOUBLE ---------------> +------------------+
831 +------------------+ | codeword |
832 | addr of DOUBLE | +------------------+
833 +------------------+ | addr of DUP |
834 | addr of EXIT | +------------------+
835 +------------------+ | addr of + |
837 %esi -> | addr of EXIT |
840 What happens when the + function does NEXT? Well, the following code is executed.
843 defcode "EXIT",4,,EXIT
844 POPRSP %esi // pop return stack into %esi
848 EXIT gets the old %esi which we saved from before on the return stack, and puts it in %esi.
849 So after this (but just before NEXT) we get:
854 +------------------+ DOUBLE
855 | addr of DOUBLE ---------------> +------------------+
856 +------------------+ | codeword |
857 %esi -> | addr of DOUBLE | +------------------+
858 +------------------+ | addr of DUP |
859 | addr of EXIT | +------------------+
860 +------------------+ | addr of + |
865 And NEXT just completes the job by, well in this case just by calling DOUBLE again :-)
867 LITERALS ----------------------------------------------------------------------
869 The final point I "glossed over" before was how to deal with functions that do anything
870 apart from calling other functions. For example, suppose that DOUBLE was defined like this:
874 It does the same thing, but how do we compile it since it contains the literal 2? One way
875 would be to have a function called "2" (which you'd have to write in assembler), but you'd need
876 a function for every single literal that you wanted to use.
878 FORTH solves this by compiling the function using a special word called LIT:
880 +---------------------------+-------+-------+-------+-------+-------+
881 | (usual header of DOUBLE) | DOCOL | LIT | 2 | * | EXIT |
882 +---------------------------+-------+-------+-------+-------+-------+
884 LIT is executed in the normal way, but what it does next is definitely not normal. It
885 looks at %esi (which now points to the literal 2), grabs it, pushes it on the stack, then
886 manipulates %esi in order to skip the literal as if it had never been there.
888 What's neat is that the whole grab/manipulate can be done using a single byte single
889 i386 instruction, our old friend LODSL. Rather than me drawing more ASCII-art diagrams,
890 see if you can find out how LIT works:
894 // %esi points to the next command, but in this case it points to the next
895 // literal 32 bit integer. Get that literal into %eax and increment %esi.
896 // On x86, it's a convenient single byte instruction! (cf. NEXT macro)
898 push %eax // push the literal number on to stack
902 MEMORY ----------------------------------------------------------------------
904 As important point about FORTH is that it gives you direct access to the lowest levels
905 of the machine. Manipulating memory directly is done frequently in FORTH, and these are
906 the primitive words for doing it.
910 pop %ebx // address to store at
911 pop %eax // data to store there
912 mov %eax,(%ebx) // store it
916 pop %ebx // address to fetch
917 mov (%ebx),%eax // fetch it
918 push %eax // push value onto stack
921 defcode "+!",2,,ADDSTORE
923 pop %eax // the amount to add
924 addl %eax,(%ebx) // add it
927 defcode "-!",2,,SUBSTORE
929 pop %eax // the amount to subtract
930 subl %eax,(%ebx) // add it
933 /* ! and @ (STORE and FETCH) store 32-bit words. It's also useful to be able to read and write bytes.
934 * I don't know whether FORTH has these words, so I invented my own, called !b and @b.
935 * Byte-oriented operations only work on architectures which permit them (i386 is one of those).
936 * UPDATE: writing a byte to the dictionary pointer is called C, in FORTH.
938 defcode "!b",2,,STOREBYTE
939 pop %ebx // address to store at
940 pop %eax // data to store there
941 movb %al,(%ebx) // store it
944 defcode "@b",2,,FETCHBYTE
945 pop %ebx // address to fetch
947 movb (%ebx),%al // fetch it
948 push %eax // push value onto stack
952 BUILT-IN VARIABLES ----------------------------------------------------------------------
954 These are some built-in variables and related standard FORTH words. Of these, the only one that we
955 have discussed so far was LATEST, which points to the last (most recently defined) word in the
956 FORTH dictionary. LATEST is also a FORTH word which pushes the address of LATEST (the variable)
957 on to the stack, so you can read or write it using @ and ! operators. For example, to print
958 the current value of LATEST (and this can apply to any FORTH variable) you would do:
962 To make defining variables shorter, I'm using a macro called defvar, similar to defword and
963 defcode above. (In fact the defvar macro uses defcode to do the dictionary header).
966 .macro defvar name, namelen, flags=0, label, initial=0
967 defcode \name,\namelen,\flags,\label
977 The built-in variables are:
979 STATE Is the interpreter executing code (0) or compiling a word (non-zero)?
980 LATEST Points to the latest (most recently defined) word in the dictionary.
981 HERE Points to the next free byte of memory. When compiling, compiled words go here.
982 _X These are three scratch variables, used by some standard dictionary words.
985 S0 Stores the address of the top of the parameter stack.
986 R0 Stores the address of the top of the return stack.
989 defvar "STATE",5,,STATE
990 defvar "HERE",4,,HERE,user_defs_start
991 defvar "LATEST",6,,LATEST,name_SYSEXIT // SYSEXIT must be last in built-in dictionary
996 defvar "R0",2,,RZ,return_stack
999 RETURN STACK ----------------------------------------------------------------------
1001 These words allow you to access the return stack. Recall that the register %ebp always points to
1002 the top of the return stack.
1006 pop %eax // pop parameter stack into %eax
1007 PUSHRSP %eax // push it on to the return stack
1010 defcode "R>",2,,FROMR
1011 POPRSP %eax // pop return stack on to %eax
1012 push %eax // and push on to parameter stack
1015 defcode "RSP@",4,,RSPFETCH
1019 defcode "RSP!",4,,RSPSTORE
1023 defcode "RDROP",5,,RDROP
1024 lea 4(%ebp),%ebp // pop return stack and throw away
1028 PARAMETER (DATA) STACK ----------------------------------------------------------------------
1030 These functions allow you to manipulate the parameter stack. Recall that Linux sets up the parameter
1031 stack for us, and it is accessed through %esp.
1034 defcode "DSP@",4,,DSPFETCH
1039 defcode "DSP!",4,,DSPSTORE
1044 INPUT AND OUTPUT ----------------------------------------------------------------------
1046 These are our first really meaty/complicated FORTH primitives. I have chosen to write them in
1047 assembler, but surprisingly in "real" FORTH implementations these are often written in terms
1048 of more fundamental FORTH primitives. I chose to avoid that because I think that just obscures
1049 the implementation. After all, you may not understand assembler but you can just think of it
1050 as an opaque block of code that does what it says.
1052 Let's discuss input first.
1054 The FORTH word KEY reads the next byte from stdin (and pushes it on the parameter stack).
1055 So if KEY is called and someone hits the space key, then the number 32 (ASCII code of space)
1056 is pushed on the stack.
1058 In FORTH there is no distinction between reading code and reading input. We might be reading
1059 and compiling code, we might be reading words to execute, we might be asking for the user
1060 to type their name -- ultimately it all comes in through KEY.
1062 The implementation of KEY uses an input buffer so a certain size (defined at the end of the
1063 program). It calls the Linux read(2) system call to fill this buffer and tracks its position
1064 in the buffer using a couple of variables, and if it runs out of input buffer then it refills
1065 it automatically. The other thing that KEY does is if it detects that stdin has closed, it
1066 exits the program, which is why when you hit ^D the FORTH system cleanly exits.
1069 #include <asm-i386/unistd.h>
1071 defcode "KEY",3,,KEY
1073 push %eax // push return value on stack
1085 1: // out of input; use read(2) to fetch more input from stdin
1086 xor %ebx,%ebx // 1st param: stdin
1087 mov $buffer,%ecx // 2nd param: buffer
1089 mov $buffend-buffer,%edx // 3rd param: max length
1090 mov $__NR_read,%eax // syscall: read
1092 test %eax,%eax // If %eax <= 0, then exit.
1094 addl %eax,%ecx // buffer+%eax = bufftop
1098 2: // error or out of input: exit
1100 mov $__NR_exit,%eax // syscall: exit
1104 By contrast, output is much simpler. The FORTH word EMIT writes out a single byte to stdout.
1105 This implementation just uses the write system call. No attempt is made to buffer output, but
1106 it would be a good exercise to add it.
1109 defcode "EMIT",4,,EMIT
1114 mov $1,%ebx // 1st param: stdout
1116 // write needs the address of the byte to write
1118 mov $2f,%ecx // 2nd param: address
1120 mov $1,%edx // 3rd param: nbytes = 1
1122 mov $__NR_write,%eax // write syscall
1127 2: .space 1 // scratch used by EMIT
1130 Back to input, WORD is a FORTH word which reads the next full word of input.
1132 What it does in detail is that it first skips any blanks (spaces, tabs, newlines and so on).
1133 Then it calls KEY to read characters into an internal buffer until it hits a blank. Then it
1134 calculates the length of the word it read and returns the address and the length as
1135 two words on the stack (with address at the top).
1137 Notice that WORD has a single internal buffer which it overwrites each time (rather like
1138 a static C string). Also notice that WORD's internal buffer is just 32 bytes long and
1139 there is NO checking for overflow. 31 bytes happens to be the maximum length of a
1140 FORTH word that we support, and that is what WORD is used for: to read FORTH words when
1141 we are compiling and executing code. The returned strings are not NUL-terminated, so
1142 in some crazy-world you could define FORTH words containing ASCII NULs, although why
1143 you'd want to is a bit beyond me.
1145 WORD is not suitable for just reading strings (eg. user input) because of all the above
1146 peculiarities and limitations.
1148 Note that when executing, you'll see:
1150 which puts "FOO" and length 3 on the stack, but when compiling:
1152 is an error (or at least it doesn't do what you might expect). Later we'll talk about compiling
1153 and immediate mode, and you'll understand why.
1156 defcode "WORD",4,,WORD
1158 push %ecx // push length
1159 push %edi // push base address
1163 /* Search for first non-blank character. Also skip \ comments. */
1165 call _KEY // get next key, returned in %eax
1166 cmpb $'\\',%al // start of a comment?
1167 je 3f // if so, skip the comment
1169 jbe 1b // if so, keep looking
1171 /* Search for the end of the word, storing chars as we go. */
1172 mov $5f,%edi // pointer to return buffer
1174 stosb // add character to return buffer
1175 call _KEY // get next key, returned in %al
1176 cmpb $' ',%al // is blank?
1177 ja 2b // if not, keep looping
1179 /* Return the word (well, the static buffer) and length. */
1181 mov %edi,%ecx // return length of the word
1182 mov $5f,%edi // return address of the word
1185 /* Code to skip \ comments to end of the current line. */
1188 cmpb $'\n',%al // end of line yet?
1193 // A static buffer where WORD returns. Subsequent calls
1194 // overwrite this buffer. Maximum word length is 32 chars.
1198 . (also called DOT) prints the top of the stack as an integer. In real FORTH implementations
1199 it should print it in the current base, but this assembler version is simpler and can only
1202 Remember that you can override even built-in FORTH words easily, so if you want to write a
1203 more advanced DOT then you can do so easily at a later point, and probably in FORTH.
1207 pop %eax // Get the number to print into %eax
1208 call _DOT // Easier to do this recursively ...
1211 mov $10,%ecx // Base 10
1215 xor %edx,%edx // %edx:%eax / %ecx -> quotient %eax, remainder %edx
1230 Almost the opposite of DOT (but not quite), SNUMBER parses a numeric string such as one returned
1231 by WORD and pushes the number on the parameter stack.
1233 This function does absolutely no error checking, and in particular the length of the string
1234 must be >= 1 bytes, and should contain only digits 0-9. If it doesn't you'll get random results.
1236 This function is only used when reading literal numbers in code, and shouldn't really be used
1237 in user code at all.
1239 defcode "SNUMBER",7,,SNUMBER
1249 imull $10,%eax // %eax *= 10
1252 subb $'0',%bl // ASCII -> digit
1259 DICTIONARY LOOK UPS ----------------------------------------------------------------------
1261 We're building up to our prelude on how FORTH code is compiled, but first we need yet more infrastructure.
1263 The FORTH word FIND takes a string (a word as parsed by WORD -- see above) and looks it up in the
1264 dictionary. What it actually returns is the address of the dictionary header, if it finds it,
1267 So if DOUBLE is defined in the dictionary, then WORD DOUBLE FIND returns the following pointer:
1273 +---------+---+---+---+---+---+---+---+---+------------+------------+------------+------------+
1274 | LINK | 6 | D | O | U | B | L | E | 0 | DOCOL | DUP | + | EXIT |
1275 +---------+---+---+---+---+---+---+---+---+------------+------------+------------+------------+
1277 See also >CFA which takes a dictionary entry pointer and returns a pointer to the codeword.
1279 FIND doesn't find dictionary entries which are flagged as HIDDEN. See below for why.
1282 defcode "FIND",4,,FIND
1283 pop %edi // %edi = address
1284 pop %ecx // %ecx = length
1290 push %esi // Save %esi so we can use it in string comparison.
1292 // Now we start searching backwards through the dictionary for this word.
1293 mov var_LATEST,%edx // LATEST points to name header of the latest word in the dictionary
1295 test %edx,%edx // NULL pointer? (end of the linked list)
1298 // Compare the length expected and the length of the word.
1299 // Note that if the F_HIDDEN flag is set on the word, then by a bit of trickery
1300 // this won't pick the word (the length will appear to be wrong).
1302 movb 4(%edx),%al // %al = flags+length field
1303 andb $(F_HIDDEN|0x1f),%al // %al = name length
1304 cmpb %cl,%al // Length is the same?
1307 // Compare the strings in detail.
1308 push %ecx // Save the length
1309 push %edi // Save the address (repe cmpsb will move this pointer)
1310 lea 5(%edx),%esi // Dictionary string we are checking against.
1311 repe cmpsb // Compare the strings.
1314 jne 2f // Not the same.
1316 // The strings are the same - return the header pointer in %eax
1322 mov (%edx),%edx // Move back through the link field to the previous word
1323 jmp 1b // .. and loop.
1327 xor %eax,%eax // Return zero to indicate not found.
1331 FIND returns the dictionary pointer, but when compiling we need the codeword pointer (recall
1332 that FORTH definitions are compiled into lists of codeword pointers).
1334 In the example below, WORD DOUBLE FIND >CFA
1336 FIND returns a pointer to this
1337 | >CFA converts it to a pointer to this
1340 +---------+---+---+---+---+---+---+---+---+------------+------------+------------+------------+
1341 | LINK | 6 | D | O | U | B | L | E | 0 | DOCOL | DUP | + | EXIT |
1342 +---------+---+---+---+---+---+---+---+---+------------+------------+------------+------------+
1346 Because names vary in length, this isn't just a simple increment.
1348 In this FORTH you cannot easily turn a codeword pointer back into a dictionary entry pointer, but
1349 that is not true in most FORTH implementations where they store a back pointer in the definition
1350 (with an obvious memory/complexity cost). The reason they do this is that it is useful to be
1351 able to go backwards (codeword -> dictionary entry) in order to decompile FORTH definitions.
1354 defcode ">CFA",4,,TCFA
1361 add $4,%edi // Skip link pointer.
1362 movb (%edi),%al // Load flags+len into %al.
1363 inc %edi // Skip flags+len byte.
1364 andb $0x1f,%al // Just the length, not the flags.
1365 add %eax,%edi // Skip the name.
1366 addl $3,%edi // The codeword is 4-byte aligned.
1371 COMPILING ----------------------------------------------------------------------
1373 Now we'll talk about how FORTH compiles words. Recall that a word definition looks like this:
1377 and we have to turn this into:
1379 pointer to previous word
1382 +--|------+---+---+---+---+---+---+---+---+------------+------------+------------+------------+
1383 | LINK | 6 | D | O | U | B | L | E | 0 | DOCOL | DUP | + | EXIT |
1384 +---------+---+---+---+---+---+---+---+---+------------+--|---------+------------+------------+
1385 ^ len pad codeword |
1387 LATEST points here points to codeword of DUP
1389 There are several problems to solve. Where to put the new word? How do we read words? How
1390 do we define : (COLON) and ; (SEMICOLON)?
1392 FORTH solves this rather elegantly and as you might expect in a very low-level way which
1393 allows you to change how the compiler works in your own code.
1395 FORTH has an INTERPRETER function (a true interpreter this time, not DOCOL) which runs in a
1396 loop, reading words (using WORD), looking them up (using FIND), turning them into codeword
1397 points (using >CFA) and deciding what to do with them. What it does depends on the mode
1398 of the interpreter (in variable STATE). When STATE is zero, the interpreter just runs
1399 each word as it looks them up. (Known as immediate mode).
1401 The interesting stuff happens when STATE is non-zero -- compiling mode. In this mode the
1402 interpreter just appends the codeword pointers to user memory (the HERE variable points to
1403 the next free byte of user memory).
1405 So you may be able to see how we could define : (COLON). The general plan is:
1407 (1) Use WORD to read the name of the function being defined.
1409 (2) Construct the dictionary entry header in user memory:
1411 pointer to previous word (from LATEST) +-- Afterwards, HERE points here, where
1412 ^ | the interpreter will start appending
1414 +--|------+---+---+---+---+---+---+---+---+------------+
1415 | LINK | 6 | D | O | U | B | L | E | 0 | DOCOL |
1416 +---------+---+---+---+---+---+---+---+---+------------+
1419 (3) Set LATEST to point to the newly defined word and most importantly leave HERE pointing
1420 just after the new codeword. This is where the interpreter will append codewords.
1422 (4) Set STATE to 1. Go into compile mode so the interpreter starts appending codewords.
1424 After : has run, our input is here:
1429 Next byte returned by KEY
1431 so the interpreter (now it's in compile mode, so I guess it's really the compiler) reads DUP,
1432 gets its codeword pointer, and appends it:
1434 +-- HERE updated to point here.
1437 +---------+---+---+---+---+---+---+---+---+------------+------------+
1438 | LINK | 6 | D | O | U | B | L | E | 0 | DOCOL | DUP |
1439 +---------+---+---+---+---+---+---+---+---+------------+------------+
1442 Next we read +, get the codeword pointer, and append it:
1444 +-- HERE updated to point here.
1447 +---------+---+---+---+---+---+---+---+---+------------+------------+------------+
1448 | LINK | 6 | D | O | U | B | L | E | 0 | DOCOL | DUP | + |
1449 +---------+---+---+---+---+---+---+---+---+------------+------------+------------+
1452 The issue is what happens next. Obviously what we _don't_ want to happen is that we
1453 read ; and compile it and go on compiling everything afterwards.
1455 At this point, FORTH uses a trick. Remember the length byte in the dictionary definition
1456 isn't just a plain length byte, but can also contain flags. One flag is called the
1457 IMMEDIATE flag (F_IMMED in this code). If a word in the dictionary is flagged as
1458 IMMEDIATE then the interpreter runs it immediately _even if it's in compile mode_.
1460 I hope I don't need to explain that ; (SEMICOLON) is an IMMEDIATE flagged word. And
1461 all it does is append the codeword for EXIT on to the current definition and switch
1462 back to immediate mode (set STATE back to 0). After executing ; we get this:
1464 +---------+---+---+---+---+---+---+---+---+------------+------------+------------+------------+
1465 | LINK | 6 | D | O | U | B | L | E | 0 | DOCOL | DUP | + | EXIT |
1466 +---------+---+---+---+---+---+---+---+---+------------+------------+------------+------------+
1471 And that's it, job done, our new definition is compiled.
1473 The only last wrinkle in this is that while our word was being compiled, it was in a
1474 half-finished state. We certainly wouldn't want DOUBLE to be called somehow during
1475 this time. There are several ways to stop this from happening, but in FORTH what we
1476 do is flag the word with the HIDDEN flag (F_HIDDEN in this code) just while it is
1477 being compiled. This prevents FIND from finding it, and thus in theory stops any
1478 chance of it being called.
1480 Compared to the description above, the actual definition of : (COLON) is comparatively simple:
1483 defcode ":",1,,COLON
1485 // Get the word and create a dictionary entry header for it.
1486 call _WORD // Returns %ecx = length, %edi = pointer to word.
1487 mov %edi,%ebx // %ebx = address of the word
1489 movl var_HERE,%edi // %edi is the address of the header
1490 movl var_LATEST,%eax // Get link pointer
1491 stosl // and store it in the header.
1493 mov %cl,%al // Get the length.
1494 orb $F_HIDDEN,%al // Set the HIDDEN flag on this entry.
1495 stosb // Store the length/flags byte.
1497 mov %ebx,%esi // %esi = word
1498 rep movsb // Copy the word
1500 addl $3,%edi // Align to next 4 byte boundary.
1503 movl $DOCOL,%eax // The codeword for user-created words is always DOCOL (the interpreter)
1506 // Header built, so now update LATEST and HERE.
1507 // We'll be compiling words and putting them HERE.
1509 movl %eax,var_LATEST
1512 // And go into compile mode by setting STATE to 1.
1517 , (COMMA) is a standard FORTH word which appends a 32 bit integer (normally a codeword
1518 pointer) to the user data area pointed to by HERE, and adds 4 to HERE.
1521 defcode ",",1,,COMMA
1522 pop %eax // Code pointer to store.
1526 movl var_HERE,%edi // HERE
1528 movl %edi,var_HERE // Update HERE (incremented)
1532 ; (SEMICOLON) is also elegantly simple. Notice the F_IMMED flag.
1535 defcode ";",1,F_IMMED,SEMICOLON
1536 movl $EXIT,%eax // EXIT is the final codeword in compiled words.
1537 call _COMMA // Store it.
1538 call _HIDDEN // Toggle the HIDDEN flag (unhides the new word).
1539 xor %eax,%eax // Set STATE to 0 (back to execute mode).
1544 IMMEDIATE mode words aren't just for the FORTH compiler to use. You can define your
1545 own IMMEDIATE words too. The IMMEDIATE word toggles the F_IMMED (IMMEDIATE flag) on the
1546 most recently defined word, or on the current word if you call it in the middle of a
1551 : MYIMMEDWORD IMMEDIATE
1555 but some FORTH programmers write this instead:
1561 The two are basically equivalent.
1564 defcode "IMMEDIATE",9,F_IMMED,IMMEDIATE
1568 movl var_LATEST,%edi // LATEST word.
1569 addl $4,%edi // Point to name/flags byte.
1570 xorb $F_IMMED,(%edi) // Toggle the IMMED bit.
1574 HIDDEN toggles the other flag, F_HIDDEN, of the latest word. Note that words flagged
1575 as hidden are defined but cannot be called, so this is rarely used.
1578 defcode "HIDDEN",6,,HIDDEN
1582 movl var_LATEST,%edi // LATEST word.
1583 addl $4,%edi // Point to name/flags byte.
1584 xorb $F_HIDDEN,(%edi) // Toggle the HIDDEN bit.
1588 ' (TICK) is a standard FORTH word which returns the codeword pointer of the next word.
1590 The common usage is:
1594 which appends the codeword of FOO to the current word we are defining (this only works in compiled code).
1596 You tend to use ' in IMMEDIATE words. For example an alternate (and rather useless) way to define
1597 a literal 2 might be:
1600 ' LIT , \ Appends LIT to the currently-being-defined word
1601 2 , \ Appends the number 2 to the currently-being-defined word
1608 (If you don't understand how LIT2 works, then you should review the material about compiling words
1609 and immediate mode).
1611 This definition of ' uses a cheat which I copied from buzzard92. As a result it only works in
1615 lodsl // Get the address of the next word and skip it.
1616 pushl %eax // Push it on the stack.
1620 BRANCHING ----------------------------------------------------------------------
1622 It turns out that all you need in order to define looping constructs, IF-statements, etc.
1625 BRANCH is an unconditional branch. 0BRANCH is a conditional branch (it only branches if the
1626 top of stack is zero).
1628 This is how BRANCH works. When BRANCH executes, %esi starts by pointing to the offset:
1630 +---------------------+-------+---- - - ---+------------+------------+---- - - - ----+------------+
1631 | (Dictionary header) | DOCOL | | BRANCH | offset | (skipped) | word |
1632 +---------------------+-------+---- - - ---+------------+-----|------+---- - - - ----+------------+
1635 | +-----------------------+
1636 %esi added to offset
1638 The offset is added to %esi to make the new %esi, and the result is that when NEXT runs, execution
1639 continues at the branch target. Negative offsets work as expected.
1641 0BRANCH is the same except the branch happens conditionally.
1643 Now standard FORTH words such as IF, THEN, ELSE, WHILE, REPEAT, etc. are implemented entirely
1644 in FORTH. They are IMMEDIATE words which append various combinations of BRANCH or 0BRANCH
1645 into the word currently being compiled.
1647 As an example, code written like this:
1649 condition-code IF true-part THEN rest-code
1653 condition-code 0BRANCH OFFSET true-part rest-code
1659 defcode "BRANCH",6,,BRANCH
1660 add (%esi),%esi // add the offset to the instruction pointer
1663 defcode "0BRANCH",7,,ZBRANCH
1665 test %eax,%eax // top of stack is zero?
1666 jz code_BRANCH // if so, jump back to the branch function above
1667 lodsl // otherwise we need to skip the offset
1671 PRINTING STRINGS ----------------------------------------------------------------------
1673 LITSTRING and EMITSTRING are primitives used to implement the ." operator (which is
1674 written in FORTH). See the definition of that operator below.
1677 defcode "LITSTRING",9,,LITSTRING
1678 lodsl // get the length of the string
1679 push %eax // push it on the stack
1680 push %esi // push the address of the start of the string
1681 addl %eax,%esi // skip past the string
1682 addl $3,%esi // but round up to next 4 byte boundary
1686 defcode "EMITSTRING",10,,EMITSTRING
1687 mov $1,%ebx // 1st param: stdout
1688 pop %ecx // 2nd param: address of string
1689 pop %edx // 3rd param: length of string
1690 mov $__NR_write,%eax // write syscall
1695 COLD START AND INTERPRETER ----------------------------------------------------------------------
1697 COLD is the first FORTH function called, almost immediately after the FORTH system "boots".
1699 INTERPRETER is the FORTH interpreter ("toploop", "toplevel" or REPL might be a more accurate
1704 // COLD must not return (ie. must not call EXIT).
1705 defword "COLD",4,,COLD
1706 .int INTERPRETER // call the interpreter loop (never returns)
1707 .int LIT,1,SYSEXIT // hmmm, but in case it does, exit(1).
1709 /* This interpreter is pretty simple, but remember that in FORTH you can always override
1710 * it later with a more powerful one!
1712 defword "INTERPRETER",11,,INTERPRETER
1713 .int INTERPRET,RDROP,INTERPRETER
1715 defcode "INTERPRET",9,,INTERPRET
1716 call _WORD // Returns %ecx = length, %edi = pointer to word.
1718 // Is it in the dictionary?
1720 movl %eax,interpret_is_lit // Not a literal number (not yet anyway ...)
1721 call _FIND // Returns %eax = pointer to header or 0 if not found.
1722 test %eax,%eax // Found?
1725 // In the dictionary. Is it an IMMEDIATE codeword?
1726 mov %eax,%edi // %edi = dictionary entry
1727 movb 4(%edi),%al // Get name+flags.
1728 push %ax // Just save it for now.
1729 call _TCFA // Convert dictionary entry (in %edi) to codeword pointer.
1731 andb $F_IMMED,%al // Is IMMED flag set?
1733 jnz 4f // If IMMED, jump straight to executing.
1737 1: // Not in the dictionary (not a word) so assume it's a literal number.
1738 incl interpret_is_lit
1739 call _SNUMBER // Returns the parsed number in %eax
1741 mov $LIT,%eax // The word is LIT
1743 2: // Are we compiling or executing?
1746 jz 4f // Jump if executing.
1748 // Compiling - just append the word to the current dictionary definition.
1750 mov interpret_is_lit,%ecx // Was it a literal?
1753 mov %ebx,%eax // Yes, so LIT is followed by a number.
1757 4: // Executing - run it!
1758 mov interpret_is_lit,%ecx // Literal?
1759 test %ecx,%ecx // Literal?
1762 // Not a literal, execute it now. This never returns, but the codeword will
1763 // eventually call NEXT which will reenter the loop in INTERPRETER.
1766 5: // Executing a literal, which means push it on the stack.
1773 .int 0 // Flag used to record if reading a literal
1776 ODDS AND ENDS ----------------------------------------------------------------------
1778 CHAR puts the ASCII code of the first character of the following word on the stack. For example
1779 CHAR A puts 65 on the stack.
1781 SYSEXIT pops the status off the stack and exits the process (using Linux exit syscall).
1784 defcode "CHAR",4,,CHAR
1785 call _WORD // Returns %ecx = length, %edi = pointer to word.
1787 movb (%edi),%al // Get the first character of the word.
1788 push %eax // Push it onto the stack.
1791 // NB: SYSEXIT must be the last entry in the built-in dictionary.
1792 defcode SYSEXIT,7,,SYSEXIT
1798 START OF FORTH CODE ----------------------------------------------------------------------
1800 We've now reached the stage where the FORTH system is running and self-hosting. All further
1801 words can be written as FORTH itself, including words like IF, THEN, .", etc which in most
1802 languages would be considered rather fundamental.
1804 As a kind of trick, I prefill the input buffer with the initial FORTH code. Once this code
1805 has run (when we get to the "OK" prompt), this input buffer is reused for reading user input.
1807 Some notes about the code:
1809 \ (backslash) is the FORTH way to start a comment which goes up to the next newline. However
1810 because this is a C-style string, I have to escape the backslash, which is why they appear as
1813 Similarly, any backslashes in the code are doubled, and " becomes \" (eg. the definition of ."
1814 is written as : .\" ... ;)
1816 I use indenting to show structure. The amount of whitespace has no meaning to FORTH however
1817 except that you must use at least one whitespace character between words, and words themselves
1818 cannot contain whitespace.
1820 FORTH is case-sensitive. Use capslock!
1828 // Multi-line constant gives 'Warning: unterminated string; newline inserted' messages which you can ignore
1830 \\ Define some character constants
1836 \\ CR prints a carriage return
1839 \\ SPACE prints a space
1840 : SPACE 'SPACE' EMIT ;
1842 \\ Primitive . (DOT) function doesn't follow with a blank, so redefine it to behave like FORTH.
1843 \\ Notice how we can trivially redefine existing functions.
1846 \\ DUP, DROP are defined in assembly for speed, but this is how you might define them
1847 \\ in FORTH. Notice use of the scratch variables _X and _Y.
1848 \\ : DUP _X ! _X @ _X @ ;
1851 \\ The 2... versions of the standard operators work on pairs of stack entries. They're not used
1852 \\ very commonly so not really worth writing in assembler. Here is how they are defined in FORTH.
1856 \\ More standard FORTH words.
1860 \\ [ and ] allow you to break into immediate mode while compiling a word.
1861 : [ IMMEDIATE \\ define [ as an immediate word
1862 0 STATE ! \\ go into immediate mode
1866 1 STATE ! \\ go back to compile mode
1869 \\ LITERAL takes whatever is on the stack and compiles LIT <foo>
1871 ' LIT , \\ compile LIT
1872 , \\ compile the literal itself (from the stack)
1875 \\ condition IF true-part THEN rest
1877 \\ condition 0BRANCH OFFSET true-part rest
1878 \\ where OFFSET is the offset of 'rest'
1879 \\ condition IF true-part ELSE false-part THEN
1881 \\ condition 0BRANCH OFFSET true-part BRANCH OFFSET2 false-part rest
1882 \\ where OFFSET if the offset of false-part and OFFSET2 is the offset of rest
1884 \\ IF is an IMMEDIATE word which compiles 0BRANCH followed by a dummy offset, and places
1885 \\ the address of the 0BRANCH on the stack. Later when we see THEN, we pop that address
1886 \\ off the stack, calculate the offset, and back-fill the offset.
1888 ' 0BRANCH , \\ compile 0BRANCH
1889 HERE @ \\ save location of the offset on the stack
1890 0 , \\ compile a dummy offset
1895 HERE @ SWAP - \\ calculate the offset from the address saved on the stack
1896 SWAP ! \\ store the offset in the back-filled location
1900 ' BRANCH , \\ definite branch to just over the false-part
1901 HERE @ \\ save location of the offset on the stack
1902 0 , \\ compile a dummy offset
1903 SWAP \\ now back-fill the original (IF) offset
1904 DUP \\ same as for THEN word above
1909 \\ BEGIN loop-part condition UNTIL
1911 \\ loop-part condition 0BRANCH OFFSET
1912 \\ where OFFSET points back to the loop-part
1913 \\ This is like do { loop-part } while (condition) in the C language
1915 HERE @ \\ save location on the stack
1919 ' 0BRANCH , \\ compile 0BRANCH
1920 HERE @ - \\ calculate the offset from the address saved on the stack
1921 , \\ compile the offset here
1924 \\ BEGIN loop-part AGAIN
1926 \\ loop-part BRANCH OFFSET
1927 \\ where OFFSET points back to the loop-part
1928 \\ In other words, an infinite loop which can only be returned from with EXIT
1930 ' BRANCH , \\ compile BRANCH
1931 HERE @ - \\ calculate the offset back
1932 , \\ compile the offset here
1935 \\ BEGIN condition WHILE loop-part REPEAT
1937 \\ condition 0BRANCH OFFSET2 loop-part BRANCH OFFSET
1938 \\ where OFFSET points back to condition (the beginning) and OFFSET2 points to after the whole piece of code
1939 \\ So this is like a while (condition) { loop-part } loop in the C language
1941 ' 0BRANCH , \\ compile 0BRANCH
1942 HERE @ \\ save location of the offset2 on the stack
1943 0 , \\ compile a dummy offset2
1947 ' BRANCH , \\ compile BRANCH
1948 SWAP \\ get the original offset (from BEGIN)
1949 HERE @ - , \\ and compile it after BRANCH
1951 HERE @ SWAP - \\ calculate the offset2
1952 SWAP ! \\ and back-fill it in the original location
1955 \\ With the looping constructs, we can now write SPACES, which writes n spaces to stdout.
1958 SPACE \\ print a space
1959 1- \\ until we count down to 0
1964 \\ .S prints the contents of the stack. Very useful for debugging.
1966 DSP@ \\ get current stack pointer
1968 DUP @ . \\ print the stack element
1970 DUP S0 @ 4- = \\ stop when we get to the top
1975 \\ DEPTH returns the depth of the stack.
1976 : DEPTH S0 @ DSP@ - ;
1978 \\ .\" is the print string operator in FORTH. Example: .\" Something to print\"
1979 \\ The space after the operator is the ordinary space required between words.
1980 \\ This is tricky to define because it has to do different things depending on whether
1981 \\ we are compiling or in immediate mode. (Thus the word is marked IMMEDIATE so it can
1982 \\ detect this and do different things).
1983 \\ In immediate mode we just keep reading characters and printing them until we get to
1984 \\ the next double quote.
1985 \\ In compile mode we have the problem of where we're going to store the string (remember
1986 \\ that the input buffer where the string comes from may be overwritten by the time we
1987 \\ come round to running the function). We store the string in the compiled function
1989 \\ LITSTRING, string length, string rounded up to 4 bytes, EMITSTRING, ...
1991 STATE @ \\ compiling?
1993 ' LITSTRING , \\ compile LITSTRING
1994 HERE @ \\ save the address of the length word on the stack
1995 0 , \\ dummy length - we don't know what it is yet
1997 KEY \\ get next character of the string
2000 HERE @ !b \\ store the character in the compiled image
2001 1 HERE +! \\ increment HERE pointer by 1 byte
2003 DROP \\ drop the double quote character at the end
2004 DUP \\ get the saved address of the length word
2005 HERE @ SWAP - \\ calculate the length
2006 4- \\ subtract 4 (because we measured from the start of the length word)
2007 SWAP ! \\ and back-fill the length location
2008 HERE @ \\ round up to next multiple of 4 bytes for the remaining code
2012 ' EMITSTRING , \\ compile the final EMITSTRING
2014 \\ In immediate mode, just read characters and print them until we get
2015 \\ to the ending double quote. Much simpler than the above code!
2018 DUP '\"' = IF EXIT THEN
2024 \\ While compiling, [COMPILE] WORD compiles WORD if it would otherwise be IMMEDIATE.
2025 : [COMPILE] IMMEDIATE
2026 WORD \\ get the next word
2027 FIND \\ find it in the dictionary
2028 >CFA \\ get its codeword
2029 , \\ and compile that
2032 \\ RECURSE makes a recursive call to the current word that is being compiled.
2033 \\ Normally while a word is being compiled, it is marked HIDDEN so that references to the
2034 \\ same word within are calls to the previous definition of the word.
2036 LATEST @ >CFA \\ LATEST points to the word being compiled at the moment
2040 \\ ALLOT is used to allocate (static) memory when compiling. It increases HERE by
2041 \\ the amount given on the stack.
2045 \\ Finally print the welcome prompt.
2058 /* END OF jonesforth.S */