nasmjf

1 \ -*- text -*- 2 \ A sometimes minimal FORTH compiler and tutorial for Linux / i386 systems. -*- asm -*- 3 \ By Richard W.M. Jones <rich@annexia.org> http://annexia.org/forth 4 \ This is PUBLIC DOMAIN (see public domain release statement below). 5 \ $Id: jonesforth.f,v 1.18 2009-09-11 08:32:33 rich Exp $ 6 \ 7 \ The first part of this tutorial is in jonesforth.S. Get if from http://annexia.org/forth 8 \ 9 \ PUBLIC DOMAIN ---------------------------------------------------------------------- 10 \ 11 \ I, the copyright holder of this work, hereby release it into the public domain. This applies worldwide. 12 \ 13 \ In case this is not legally possible, I grant any entity the right to use this work for any purpose, 14 \ without any conditions, unless such conditions are required by law. 15 \ 16 \ SETTING UP ---------------------------------------------------------------------- 17 \ 18 \ Let's get a few housekeeping things out of the way. Firstly because I need to draw lots of 19 \ ASCII-art diagrams to explain concepts, the best way to look at this is using a window which 20 \ uses a fixed width font and is at least this wide: 21 \ 22 \<------------------------------------------------------------------------------------------------------------------------> 23 \ 24 \ Secondly make sure TABS are set to 8 characters. The following should be a vertical 25 \ line. If not, sort out your tabs. 26 \ 27 \ | 28 \ | 29 \ | 30 \ 31 \ Thirdly I assume that your screen is at least 50 characters high. 32 \ 33 \ START OF FORTH CODE ---------------------------------------------------------------------- 34 \ 35 \ We've now reached the stage where the FORTH system is running and self-hosting. All further 36 \ words can be written as FORTH itself, including words like IF, THEN, .", etc which in most 37 \ languages would be considered rather fundamental. 38 \ 39 \ Some notes about the code: 40 \ 41 \ I use indenting to show structure. The amount of whitespace has no meaning to FORTH however 42 \ except that you must use at least one whitespace character between words, and words themselves 43 \ cannot contain whitespace. 44 \ 45 \ FORTH is case-sensitive. Use capslock! 46 47 \ The primitive word /MOD (DIVMOD) leaves both the quotient and the remainder on the stack. (On 48 \ i386, the idivl instruction gives both anyway). Now we can define the / and MOD in terms of /MOD 49 \ and a few other primitives. 50 : / /MOD SWAP DROP ; 51 : MOD /MOD DROP ; 52 53 \ Define some character constants 54 : '\n' 10 ; 55 : BL 32 ; \ BL (BLank) is a standard FORTH word for space. 56 57 \ CR prints a carriage return 58 : CR '\n' EMIT ; 59 60 \ SPACE prints a space 61 : SPACE BL EMIT ; 62 63 \ NEGATE leaves the negative of a number on the stack. 64 : NEGATE 0 SWAP - ; 65 66 \ Standard words for booleans. 67 : TRUE 1 ; 68 : FALSE 0 ; 69 : NOT 0= ; 70 71 \ LITERAL takes whatever is on the stack and compiles LIT <foo> 72 : LITERAL IMMEDIATE 73 ' LIT , \ compile LIT 74 , \ compile the literal itself (from the stack) 75 ; 76 77 \ Now we can use [ and ] to insert literals which are calculated at compile time. (Recall that 78 \ [ and ] are the FORTH words which switch into and out of immediate mode.) 79 \ Within definitions, use [ ... ] LITERAL anywhere that '...' is a constant expression which you 80 \ would rather only compute once (at compile time, rather than calculating it each time your word runs). 81 : ':' 82 [ \ go into immediate mode (temporarily) 83 CHAR : \ push the number 58 (ASCII code of colon) on the parameter stack 84 ] \ go back to compile mode 85 LITERAL \ compile LIT 58 as the definition of ':' word 86 ; 87 88 \ A few more character constants defined the same way as above. 89 : ';' [ CHAR ; ] LITERAL ; 90 : '(' [ CHAR ( ] LITERAL ; 91 : ')' [ CHAR ) ] LITERAL ; 92 : '"' [ CHAR " ] LITERAL ; 93 : 'A' [ CHAR A ] LITERAL ; 94 : '0' [ CHAR 0 ] LITERAL ; 95 : '-' [ CHAR - ] LITERAL ; 96 : '.' [ CHAR . ] LITERAL ; 97 98 \ While compiling, '[COMPILE] word' compiles 'word' if it would otherwise be IMMEDIATE. 99 : [COMPILE] IMMEDIATE 100 WORD \ get the next word 101 FIND \ find it in the dictionary 102 >CFA \ get its codeword 103 , \ and compile that 104 ; 105 106 \ RECURSE makes a recursive call to the current word that is being compiled. 107 \ 108 \ Normally while a word is being compiled, it is marked HIDDEN so that references to the 109 \ same word within are calls to the previous definition of the word. However we still have 110 \ access to the word which we are currently compiling through the LATEST pointer so we 111 \ can use that to compile a recursive call. 112 : RECURSE IMMEDIATE 113 LATEST @ \ LATEST points to the word being compiled at the moment 114 >CFA \ get the codeword 115 , \ compile it 116 ; 117 118 \ CONTROL STRUCTURES ---------------------------------------------------------------------- 119 \ 120 \ So far we have defined only very simple definitions. Before we can go further, we really need to 121 \ make some control structures, like IF ... THEN and loops. Luckily we can define arbitrary control 122 \ structures directly in FORTH. 123 \ 124 \ Please note that the control structures as I have defined them here will only work inside compiled 125 \ words. If you try to type in expressions using IF, etc. in immediate mode, then they won't work. 126 \ Making these work in immediate mode is left as an exercise for the reader. 127 128 \ condition IF true-part THEN rest 129 \ -- compiles to: --> condition 0BRANCH OFFSET true-part rest 130 \ where OFFSET is the offset of 'rest' 131 \ condition IF true-part ELSE false-part THEN 132 \ -- compiles to: --> condition 0BRANCH OFFSET true-part BRANCH OFFSET2 false-part rest 133 \ where OFFSET if the offset of false-part and OFFSET2 is the offset of rest 134 135 \ IF is an IMMEDIATE word which compiles 0BRANCH followed by a dummy offset, and places 136 \ the address of the 0BRANCH on the stack. Later when we see THEN, we pop that address 137 \ off the stack, calculate the offset, and back-fill the offset. 138 : IF IMMEDIATE 139 ' 0BRANCH , \ compile 0BRANCH 140 HERE @ \ save location of the offset on the stack 141 0 , \ compile a dummy offset 142 ; 143 144 : THEN IMMEDIATE 145 DUP 146 HERE @ SWAP - \ calculate the offset from the address saved on the stack 147 SWAP ! \ store the offset in the back-filled location 148 ; 149 150 : ELSE IMMEDIATE 151 ' BRANCH , \ definite branch to just over the false-part 152 HERE @ \ save location of the offset on the stack 153 0 , \ compile a dummy offset 154 SWAP \ now back-fill the original (IF) offset 155 DUP \ same as for THEN word above 156 HERE @ SWAP - 157 SWAP ! 158 ; 159 160 \ BEGIN loop-part condition UNTIL 161 \ -- compiles to: --> loop-part condition 0BRANCH OFFSET 162 \ where OFFSET points back to the loop-part 163 \ This is like do { loop-part } while (condition) in the C language 164 : BEGIN IMMEDIATE 165 HERE @ \ save location on the stack 166 ; 167 168 : UNTIL IMMEDIATE 169 ' 0BRANCH , \ compile 0BRANCH 170 HERE @ - \ calculate the offset from the address saved on the stack 171 , \ compile the offset here 172 ; 173 174 \ BEGIN loop-part AGAIN 175 \ -- compiles to: --> loop-part BRANCH OFFSET 176 \ where OFFSET points back to the loop-part 177 \ In other words, an infinite loop which can only be returned from with EXIT 178 : AGAIN IMMEDIATE 179 ' BRANCH , \ compile BRANCH 180 HERE @ - \ calculate the offset back 181 , \ compile the offset here 182 ; 183 184 \ BEGIN condition WHILE loop-part REPEAT 185 \ -- compiles to: --> condition 0BRANCH OFFSET2 loop-part BRANCH OFFSET 186 \ where OFFSET points back to condition (the beginning) and OFFSET2 points to after the whole piece of code 187 \ So this is like a while (condition) { loop-part } loop in the C language 188 : WHILE IMMEDIATE 189 ' 0BRANCH , \ compile 0BRANCH 190 HERE @ \ save location of the offset2 on the stack 191 0 , \ compile a dummy offset2 192 ; 193 194 : REPEAT IMMEDIATE 195 ' BRANCH , \ compile BRANCH 196 SWAP \ get the original offset (from BEGIN) 197 HERE @ - , \ and compile it after BRANCH 198 DUP 199 HERE @ SWAP - \ calculate the offset2 200 SWAP ! \ and back-fill it in the original location 201 ; 202 203 \ UNLESS is the same as IF but the test is reversed. 204 \ 205 \ Note the use of [COMPILE]: Since IF is IMMEDIATE we don't want it to be executed while UNLESS 206 \ is compiling, but while UNLESS is running (which happens to be when whatever word using UNLESS is 207 \ being compiled -- whew!). So we use [COMPILE] to reverse the effect of marking IF as immediate. 208 \ This trick is generally used when we want to write our own control words without having to 209 \ implement them all in terms of the primitives 0BRANCH and BRANCH, but instead reusing simpler 210 \ control words like (in this instance) IF. 211 : UNLESS IMMEDIATE 212 ' NOT , \ compile NOT (to reverse the test) 213 [COMPILE] IF \ continue by calling the normal IF 214 ; 215 216 \ COMMENTS ---------------------------------------------------------------------- 217 \ 218 \ FORTH allows ( ... ) as comments within function definitions. This works by having an IMMEDIATE 219 \ word called ( which just drops input characters until it hits the corresponding ). 220 : ( IMMEDIATE 221 1 \ allowed nested parens by keeping track of depth 222 BEGIN 223 KEY \ read next character 224 DUP '(' = IF \ open paren? 225 DROP \ drop the open paren 226 1+ \ depth increases 227 ELSE 228 ')' = IF \ close paren? 229 1- \ depth decreases 230 THEN 231 THEN 232 DUP 0= UNTIL \ continue until we reach matching close paren, depth 0 233 DROP \ drop the depth counter 234 ; 235 236 ( 237 From now on we can use ( ... ) for comments. 238 239 STACK NOTATION ---------------------------------------------------------------------- 240 241 In FORTH style we can also use ( ... -- ... ) to show the effects that a word has on the 242 parameter stack. For example: 243 244 ( n -- ) means that the word consumes an integer (n) from the parameter stack. 245 ( b a -- c ) means that the word uses two integers (a and b, where a is at the top of stack) 246 and returns a single integer (c). 247 ( -- ) means the word has no effect on the stack 248 ) 249 250 ( Some more complicated stack examples, showing the stack notation. ) 251 : NIP ( x y -- y ) SWAP DROP ; 252 : TUCK ( x y -- y x y ) SWAP OVER ; 253 : PICK ( x_u ... x_1 x_0 u -- x_u ... x_1 x_0 x_u ) 254 1+ ( add one because of 'u' on the stack ) 255 4 * ( multiply by the word size ) 256 DSP@ + ( add to the stack pointer ) 257 @ ( and fetch ) 258 ; 259 260 ( With the looping constructs, we can now write SPACES, which writes n spaces to stdout. ) 261 : SPACES ( n -- ) 262 BEGIN 263 DUP 0> ( while n > 0 ) 264 WHILE 265 SPACE ( print a space ) 266 1- ( until we count down to 0 ) 267 REPEAT 268 DROP 269 ; 270 271 ( Standard words for manipulating BASE. ) 272 : DECIMAL ( -- ) 10 BASE ! ; 273 : HEX ( -- ) 16 BASE ! ; 274 275 ( 276 PRINTING NUMBERS ---------------------------------------------------------------------- 277 278 The standard FORTH word . (DOT) is very important. It takes the number at the top 279 of the stack and prints it out. However first I'm going to implement some lower-level 280 FORTH words: 281 282 U.R ( u width -- ) which prints an unsigned number, padded to a certain width 283 U. ( u -- ) which prints an unsigned number 284 .R ( n width -- ) which prints a signed number, padded to a certain width. 285 286 For example: 287 -123 6 .R 288 will print out these characters: 289 <space> <space> - 1 2 3 290 291 In other words, the number padded left to a certain number of characters. 292 293 The full number is printed even if it is wider than width, and this is what allows us to 294 define the ordinary functions U. and . (we just set width to zero knowing that the full 295 number will be printed anyway). 296 297 Another wrinkle of . and friends is that they obey the current base in the variable BASE. 298 BASE can be anything in the range 2 to 36. 299 300 While we're defining . &c we can also define .S which is a useful debugging tool. This 301 word prints the current stack (non-destructively) from top to bottom. 302 ) 303 304 ( This is the underlying recursive definition of U. ) 305 : U. ( u -- ) 306 BASE @ /MOD ( width rem quot ) 307 ?DUP IF ( if quotient <> 0 then ) 308 RECURSE ( print the quotient ) 309 THEN 310 311 ( print the remainder ) 312 DUP 10 < IF 313 '0' ( decimal digits 0..9 ) 314 ELSE 315 10 - ( hex and beyond digits A..Z ) 316 'A' 317 THEN 318 + 319 EMIT 320 ; 321 322 ( 323 FORTH word .S prints the contents of the stack. It doesn't alter the stack. 324 Very useful for debugging. 325 ) 326 : .S ( -- ) 327 DSP@ ( get current stack pointer ) 328 BEGIN 329 DUP S0 @ < 330 WHILE 331 DUP @ U. ( print the stack element ) 332 SPACE 333 4+ ( move up ) 334 REPEAT 335 DROP 336 ; 337 338 ( This word returns the width (in characters) of an unsigned number in the current base ) 339 : UWIDTH ( u -- width ) 340 BASE @ / ( rem quot ) 341 ?DUP IF ( if quotient <> 0 then ) 342 RECURSE 1+ ( return 1+recursive call ) 343 ELSE 344 1 ( return 1 ) 345 THEN 346 ; 347 348 : U.R ( u width -- ) 349 SWAP ( width u ) 350 DUP ( width u u ) 351 UWIDTH ( width u uwidth ) 352 ROT ( u uwidth width ) 353 SWAP - ( u width-uwidth ) 354 ( At this point if the requested width is narrower, we'll have a negative number on the stack. 355 Otherwise the number on the stack is the number of spaces to print. But SPACES won't print 356 a negative number of spaces anyway, so it's now safe to call SPACES ... ) 357 SPACES 358 ( ... and then call the underlying implementation of U. ) 359 U. 360 ; 361 362 ( 363 .R prints a signed number, padded to a certain width. We can't just print the sign 364 and call U.R because we want the sign to be next to the number ('-123' instead of '- 123'). 365 ) 366 : .R ( n width -- ) 367 SWAP ( width n ) 368 DUP 0< IF 369 NEGATE ( width u ) 370 1 ( save a flag to remember that it was negative | width n 1 ) 371 SWAP ( width 1 u ) 372 ROT ( 1 u width ) 373 1- ( 1 u width-1 ) 374 ELSE 375 0 ( width u 0 ) 376 SWAP ( width 0 u ) 377 ROT ( 0 u width ) 378 THEN 379 SWAP ( flag width u ) 380 DUP ( flag width u u ) 381 UWIDTH ( flag width u uwidth ) 382 ROT ( flag u uwidth width ) 383 SWAP - ( flag u width-uwidth ) 384 385 SPACES ( flag u ) 386 SWAP ( u flag ) 387 388 IF ( was it negative? print the - character ) 389 '-' EMIT 390 THEN 391 392 U. 393 ; 394 395 ( Finally we can define word . in terms of .R, with a trailing space. ) 396 : . 0 .R SPACE ; 397 398 ( The real U., note the trailing space. ) 399 : U. U. SPACE ; 400 401 ( ? fetches the integer at an address and prints it. ) 402 : ? ( addr -- ) @ . ; 403 404 ( c a b WITHIN returns true if a <= c and c < b ) 405 ( or define without ifs: OVER - >R - R> U< ) 406 : WITHIN 407 -ROT ( b c a ) 408 OVER ( b c a c ) 409 <= IF 410 > IF ( b c -- ) 411 TRUE 412 ELSE 413 FALSE 414 THEN 415 ELSE 416 2DROP ( b c -- ) 417 FALSE 418 THEN 419 ; 420 421 ( DEPTH returns the depth of the stack. ) 422 : DEPTH ( -- n ) 423 S0 @ DSP@ - 424 4- ( adjust because S0 was on the stack when we pushed DSP ) 425 ; 426 427 ( 428 ALIGNED takes an address and rounds it up (aligns it) to the next 4 byte boundary. 429 ) 430 : ALIGNED ( addr -- addr ) 431 3 + 3 INVERT AND ( (addr+3) & ~3 ) 432 ; 433 434 ( 435 ALIGN aligns the HERE pointer, so the next word appended will be aligned properly. 436 ) 437 : ALIGN HERE @ ALIGNED HERE ! ; 438 439 ( 440 STRINGS ---------------------------------------------------------------------- 441 442 S" string" is used in FORTH to define strings. It leaves the address of the string and 443 its length on the stack, (length at the top of stack). The space following S" is the normal 444 space between FORTH words and is not a part of the string. 445 446 This is tricky to define because it has to do different things depending on whether 447 we are compiling or in immediate mode. (Thus the word is marked IMMEDIATE so it can 448 detect this and do different things). 449 450 In compile mode we append 451 LITSTRING <string length> <string rounded up 4 bytes> 452 to the current word. The primitive LITSTRING does the right thing when the current 453 word is executed. 454 455 In immediate mode there isn't a particularly good place to put the string, but in this 456 case we put the string at HERE (but we _don't_ change HERE). This is meant as a temporary 457 location, likely to be overwritten soon after. 458 ) 459 ( C, appends a byte to the current compiled word. ) 460 : C, 461 HERE @ C! ( store the character in the compiled image ) 462 1 HERE +! ( increment HERE pointer by 1 byte ) 463 ; 464 465 : S" IMMEDIATE ( -- addr len ) 466 STATE @ IF ( compiling? ) 467 ' LITSTRING , ( compile LITSTRING ) 468 HERE @ ( save the address of the length word on the stack ) 469 0 , ( dummy length - we don't know what it is yet ) 470 BEGIN 471 KEY ( get next character of the string ) 472 DUP '"' <> 473 WHILE 474 C, ( copy character ) 475 REPEAT 476 DROP ( drop the double quote character at the end ) 477 DUP ( get the saved address of the length word ) 478 HERE @ SWAP - ( calculate the length ) 479 4- ( subtract 4 (because we measured from the start of the length word) ) 480 SWAP ! ( and back-fill the length location ) 481 ALIGN ( round up to next multiple of 4 bytes for the remaining code ) 482 ELSE ( immediate mode ) 483 HERE @ ( get the start address of the temporary space ) 484 BEGIN 485 KEY 486 DUP '"' <> 487 WHILE 488 OVER C! ( save next character ) 489 1+ ( increment address ) 490 REPEAT 491 DROP ( drop the final " character ) 492 HERE @ - ( calculate the length ) 493 HERE @ ( push the start address ) 494 SWAP ( addr len ) 495 THEN 496 ; 497 498 ( 499 ." is the print string operator in FORTH. Example: ." Something to print" 500 The space after the operator is the ordinary space required between words and is not 501 a part of what is printed. 502 503 In immediate mode we just keep reading characters and printing them until we get to 504 the next double quote. 505 506 In compile mode we use S" to store the string, then add TELL afterwards: 507 LITSTRING <string length> <string rounded up to 4 bytes> TELL 508 509 It may be interesting to note the use of [COMPILE] to turn the call to the immediate 510 word S" into compilation of that word. It compiles it into the definition of .", 511 not into the definition of the word being compiled when this is running (complicated 512 enough for you?) 513 ) 514 : ." IMMEDIATE ( -- ) 515 STATE @ IF ( compiling? ) 516 [COMPILE] S" ( read the string, and compile LITSTRING, etc. ) 517 ' TELL , ( compile the final TELL ) 518 ELSE 519 ( In immediate mode, just read characters and print them until we get 520 to the ending double quote. ) 521 BEGIN 522 KEY 523 DUP '"' = IF 524 DROP ( drop the double quote character ) 525 EXIT ( return from this function ) 526 THEN 527 EMIT 528 AGAIN 529 THEN 530 ; 531 532 ( 533 CONSTANTS AND VARIABLES ---------------------------------------------------------------------- 534 535 In FORTH, global constants and variables are defined like this: 536 537 10 CONSTANT TEN when TEN is executed, it leaves the integer 10 on the stack 538 VARIABLE VAR when VAR is executed, it leaves the address of VAR on the stack 539 540 Constants can be read but not written, eg: 541 542 TEN . CR prints 10 543 544 You can read a variable (in this example called VAR) by doing: 545 546 VAR @ leaves the value of VAR on the stack 547 VAR @ . CR prints the value of VAR 548 VAR ? CR same as above, since ? is the same as @ . 549 550 and update the variable by doing: 551 552 20 VAR ! sets VAR to 20 553 554 Note that variables are uninitialised (but see VALUE later on which provides initialised 555 variables with a slightly simpler syntax). 556 557 How can we define the words CONSTANT and VARIABLE? 558 559 The trick is to define a new word for the variable itself (eg. if the variable was called 560 'VAR' then we would define a new word called VAR). This is easy to do because we exposed 561 dictionary entry creation through the CREATE word (part of the definition of : above). 562 A call to WORD [TEN] CREATE (where [TEN] means that "TEN" is the next word in the input) 563 leaves the dictionary entry: 564 565 +--- HERE 566 | 567 V 568 +---------+---+---+---+---+ 569 | LINK | 3 | T | E | N | 570 +---------+---+---+---+---+ 571 len 572 573 For CONSTANT we can continue by appending DOCOL (the codeword), then LIT followed by 574 the constant itself and then EXIT, forming a little word definition that returns the 575 constant: 576 577 +---------+---+---+---+---+------------+------------+------------+------------+ 578 | LINK | 3 | T | E | N | DOCOL | LIT | 10 | EXIT | 579 +---------+---+---+---+---+------------+------------+------------+------------+ 580 len codeword 581 582 Notice that this word definition is exactly the same as you would have got if you had 583 written : TEN 10 ; 584 585 Note for people reading the code below: DOCOL is a constant word which we defined in the 586 assembler part which returns the value of the assembler symbol of the same name. 587 ) 588 : CONSTANT 589 WORD ( get the name (the name follows CONSTANT) ) 590 CREATE ( make the dictionary entry ) 591 DOCOL , ( append DOCOL (the codeword field of this word) ) 592 ' LIT , ( append the codeword LIT ) 593 , ( append the value on the top of the stack ) 594 ' EXIT , ( append the codeword EXIT ) 595 ; 596 597 ( 598 VARIABLE is a little bit harder because we need somewhere to put the variable. There is 599 nothing particularly special about the user memory (the area of memory pointed to by HERE 600 where we have previously just stored new word definitions). We can slice off bits of this 601 memory area to store anything we want, so one possible definition of VARIABLE might create 602 this: 603 604 +--------------------------------------------------------------+ 605 | | 606 V | 607 +---------+---------+---+---+---+---+------------+------------+---|--------+------------+ 608 | <var> | LINK | 3 | V | A | R | DOCOL | LIT | <addr var> | EXIT | 609 +---------+---------+---+---+---+---+------------+------------+------------+------------+ 610 len codeword 611 612 where <var> is the place to store the variable, and <addr var> points back to it. 613 614 To make this more general let's define a couple of words which we can use to allocate 615 arbitrary memory from the user memory. 616 617 First ALLOT, where n ALLOT allocates n bytes of memory. (Note when calling this that 618 it's a very good idea to make sure that n is a multiple of 4, or at least that next time 619 a word is compiled that HERE has been left as a multiple of 4). 620 ) 621 : ALLOT ( n -- addr ) 622 HERE @ SWAP ( here n ) 623 HERE +! ( adds n to HERE, after this the old value of HERE is still on the stack ) 624 ; 625 626 ( 627 Second, CELLS. In FORTH the phrase 'n CELLS ALLOT' means allocate n integers of whatever size 628 is the natural size for integers on this machine architecture. On this 32 bit machine therefore 629 CELLS just multiplies the top of stack by 4. 630 ) 631 : CELLS ( n -- n ) 4 * ; 632 633 ( 634 So now we can define VARIABLE easily in much the same way as CONSTANT above. Refer to the 635 diagram above to see what the word that this creates will look like. 636 ) 637 : VARIABLE 638 1 CELLS ALLOT ( allocate 1 cell of memory, push the pointer to this memory ) 639 WORD CREATE ( make the dictionary entry (the name follows VARIABLE) ) 640 DOCOL , ( append DOCOL (the codeword field of this word) ) 641 ' LIT , ( append the codeword LIT ) 642 , ( append the pointer to the new memory ) 643 ' EXIT , ( append the codeword EXIT ) 644 ; 645 646 ( 647 VALUES ---------------------------------------------------------------------- 648 649 VALUEs are like VARIABLEs but with a simpler syntax. You would generally use them when you 650 want a variable which is read often, and written infrequently. 651 652 20 VALUE VAL creates VAL with initial value 20 653 VAL pushes the value (20) directly on the stack 654 30 TO VAL updates VAL, setting it to 30 655 VAL pushes the value (30) directly on the stack 656 657 Notice that 'VAL' on its own doesn't return the address of the value, but the value itself, 658 making values simpler and more obvious to use than variables (no indirection through '@'). 659 The price is a more complicated implementation, although despite the complexity there is no 660 performance penalty at runtime. 661 662 A naive implementation of 'TO' would be quite slow, involving a dictionary search each time. 663 But because this is FORTH we have complete control of the compiler so we can compile TO more 664 efficiently, turning: 665 TO VAL 666 into: 667 LIT <addr> ! 668 and calculating <addr> (the address of the value) at compile time. 669 670 Now this is the clever bit. We'll compile our value like this: 671 672 +---------+---+---+---+---+------------+------------+------------+------------+ 673 | LINK | 3 | V | A | L | DOCOL | LIT | <value> | EXIT | 674 +---------+---+---+---+---+------------+------------+------------+------------+ 675 len codeword 676 677 where <value> is the actual value itself. Note that when VAL executes, it will push the 678 value on the stack, which is what we want. 679 680 But what will TO use for the address <addr>? Why of course a pointer to that <value>: 681 682 code compiled - - - - --+------------+------------+------------+-- - - - - 683 by TO VAL | LIT | <addr> | ! | 684 - - - - --+------------+-----|------+------------+-- - - - - 685 | 686 V 687 +---------+---+---+---+---+------------+------------+------------+------------+ 688 | LINK | 3 | V | A | L | DOCOL | LIT | <value> | EXIT | 689 +---------+---+---+---+---+------------+------------+------------+------------+ 690 len codeword 691 692 In other words, this is a kind of self-modifying code. 693 694 (Note to the people who want to modify this FORTH to add inlining: values defined this 695 way cannot be inlined). 696 ) 697 : VALUE ( n -- ) 698 WORD CREATE ( make the dictionary entry (the name follows VALUE) ) 699 DOCOL , ( append DOCOL ) 700 ' LIT , ( append the codeword LIT ) 701 , ( append the initial value ) 702 ' EXIT , ( append the codeword EXIT ) 703 ; 704 705 : TO IMMEDIATE ( n -- ) 706 WORD ( get the name of the value ) 707 FIND ( look it up in the dictionary ) 708 >DFA ( get a pointer to the first data field (the 'LIT') ) 709 4+ ( increment to point at the value ) 710 STATE @ IF ( compiling? ) 711 ' LIT , ( compile LIT ) 712 , ( compile the address of the value ) 713 ' ! , ( compile ! ) 714 ELSE ( immediate mode ) 715 ! ( update it straightaway ) 716 THEN 717 ; 718 719 ( x +TO VAL adds x to VAL ) 720 : +TO IMMEDIATE 721 WORD ( get the name of the value ) 722 FIND ( look it up in the dictionary ) 723 >DFA ( get a pointer to the first data field (the 'LIT') ) 724 4+ ( increment to point at the value ) 725 STATE @ IF ( compiling? ) 726 ' LIT , ( compile LIT ) 727 , ( compile the address of the value ) 728 ' +! , ( compile +! ) 729 ELSE ( immediate mode ) 730 +! ( update it straightaway ) 731 THEN 732 ; 733 734 ( 735 PRINTING THE DICTIONARY ---------------------------------------------------------------------- 736 737 ID. takes an address of a dictionary entry and prints the word's name. 738 739 For example: LATEST @ ID. would print the name of the last word that was defined. 740 ) 741 : ID. 742 4+ ( skip over the link pointer ) 743 DUP C@ ( get the flags/length byte ) 744 F_LENMASK AND ( mask out the flags - just want the length ) 745 746 BEGIN 747 DUP 0> ( length > 0? ) 748 WHILE 749 SWAP 1+ ( addr len -- len addr+1 ) 750 DUP C@ ( len addr -- len addr char | get the next character) 751 EMIT ( len addr char -- len addr | and print it) 752 SWAP 1- ( len addr -- addr len-1 | subtract one from length ) 753 REPEAT 754 2DROP ( len addr -- ) 755 ; 756 757 ( 758 'WORD word FIND ?HIDDEN' returns true if 'word' is flagged as hidden. 759 760 'WORD word FIND ?IMMEDIATE' returns true if 'word' is flagged as immediate. 761 ) 762 : ?HIDDEN 763 4+ ( skip over the link pointer ) 764 C@ ( get the flags/length byte ) 765 F_HIDDEN AND ( mask the F_HIDDEN flag and return it (as a truth value) ) 766 ; 767 : ?IMMEDIATE 768 4+ ( skip over the link pointer ) 769 C@ ( get the flags/length byte ) 770 F_IMMED AND ( mask the F_IMMED flag and return it (as a truth value) ) 771 ; 772 773 ( 774 WORDS prints all the words defined in the dictionary, starting with the word defined most recently. 775 However it doesn't print hidden words. 776 777 The implementation simply iterates backwards from LATEST using the link pointers. 778 ) 779 : WORDS 780 LATEST @ ( start at LATEST dictionary entry ) 781 BEGIN 782 ?DUP ( while link pointer is not null ) 783 WHILE 784 DUP ?HIDDEN NOT IF ( ignore hidden words ) 785 DUP ID. ( but if not hidden, print the word ) 786 SPACE 787 THEN 788 @ ( dereference the link pointer - go to previous word ) 789 REPEAT 790 CR 791 ; 792 793 ( 794 FORGET ---------------------------------------------------------------------- 795 796 So far we have only allocated words and memory. FORTH provides a rather primitive method 797 to deallocate. 798 799 'FORGET word' deletes the definition of 'word' from the dictionary and everything defined 800 after it, including any variables and other memory allocated after. 801 802 The implementation is very simple - we look up the word (which returns the dictionary entry 803 address). Then we set HERE to point to that address, so in effect all future allocations 804 and definitions will overwrite memory starting at the word. We also need to set LATEST to 805 point to the previous word. 806 807 Note that you cannot FORGET built-in words (well, you can try but it will probably cause 808 a segfault). 809 810 XXX: Because we wrote VARIABLE to store the variable in memory allocated before the word, 811 in the current implementation VARIABLE FOO FORGET FOO will leak 1 cell of memory. 812 ) 813 : FORGET 814 WORD FIND ( find the word, gets the dictionary entry address ) 815 DUP @ LATEST ! ( set LATEST to point to the previous word ) 816 HERE ! ( and store HERE with the dictionary address ) 817 ; 818 819 ( 820 DUMP ---------------------------------------------------------------------- 821 822 DUMP is used to dump out the contents of memory, in the 'traditional' hexdump format. 823 824 Notice that the parameters to DUMP (address, length) are compatible with string words 825 such as WORD and S". 826 827 You can dump out the raw code for the last word you defined by doing something like: 828 829 LATEST @ 128 DUMP 830 ) 831 : DUMP ( addr len -- ) 832 BASE @ -ROT ( save the current BASE at the bottom of the stack ) 833 HEX ( and switch to hexadecimal mode ) 834 835 BEGIN 836 ?DUP ( while len > 0 ) 837 WHILE 838 OVER 8 U.R ( print the address ) 839 SPACE 840 841 ( print up to 16 words on this line ) 842 2DUP ( addr len addr len ) 843 1- 15 AND 1+ ( addr len addr linelen ) 844 BEGIN 845 ?DUP ( while linelen > 0 ) 846 WHILE 847 SWAP ( addr len linelen addr ) 848 DUP C@ ( addr len linelen addr byte ) 849 2 .R SPACE ( print the byte ) 850 1+ SWAP 1- ( addr len linelen addr -- addr len addr+1 linelen-1 ) 851 REPEAT 852 DROP ( addr len ) 853 854 ( print the ASCII equivalents ) 855 2DUP 1- 15 AND 1+ ( addr len addr linelen ) 856 BEGIN 857 ?DUP ( while linelen > 0) 858 WHILE 859 SWAP ( addr len linelen addr ) 860 DUP C@ ( addr len linelen addr byte ) 861 DUP 32 128 WITHIN IF ( 32 <= c < 128? ) 862 EMIT 863 ELSE 864 DROP '.' EMIT 865 THEN 866 1+ SWAP 1- ( addr len linelen addr -- addr len addr+1 linelen-1 ) 867 REPEAT 868 DROP ( addr len ) 869 CR 870 871 DUP 1- 15 AND 1+ ( addr len linelen ) 872 TUCK ( addr linelen len linelen ) 873 - ( addr linelen len-linelen ) 874 >R + R> ( addr+linelen len-linelen ) 875 REPEAT 876 877 DROP ( restore stack ) 878 BASE ! ( restore saved BASE ) 879 ; 880 881 ( 882 CASE ---------------------------------------------------------------------- 883 884 CASE...ENDCASE is how we do switch statements in FORTH. There is no generally 885 agreed syntax for this, so I've gone for the syntax mandated by the ISO standard 886 FORTH (ANS-FORTH). 887 888 ( some value on the stack ) 889 CASE 890 test1 OF ... ENDOF 891 test2 OF ... ENDOF 892 testn OF ... ENDOF 893 ... ( default case ) 894 ENDCASE 895 896 The CASE statement tests the value on the stack by comparing it for equality with 897 test1, test2, ..., testn and executes the matching piece of code within OF ... ENDOF. 898 If none of the test values match then the default case is executed. Inside the ... of 899 the default case, the value is still at the top of stack (it is implicitly DROP-ed 900 by ENDCASE). When ENDOF is executed it jumps after ENDCASE (ie. there is no "fall-through" 901 and no need for a break statement like in C). 902 903 The default case may be omitted. In fact the tests may also be omitted so that you 904 just have a default case, although this is probably not very useful. 905 906 An example (assuming that 'q', etc. are words which push the ASCII value of the letter 907 on the stack): 908 909 0 VALUE QUIT 910 0 VALUE SLEEP 911 KEY CASE 912 'q' OF 1 TO QUIT ENDOF 913 's' OF 1 TO SLEEP ENDOF 914 ( default case: ) 915 ." Sorry, I didn't understand key <" DUP EMIT ." >, try again." CR 916 ENDCASE 917 918 (In some versions of FORTH, more advanced tests are supported, such as ranges, etc. 919 Other versions of FORTH need you to write OTHERWISE to indicate the default case. 920 As I said above, this FORTH tries to follow the ANS FORTH standard). 921 922 The implementation of CASE...ENDCASE is somewhat non-trivial. I'm following the 923 implementations from here: 924 http://www.uni-giessen.de/faq/archiv/forthfaq.case_endcase/msg00000.html 925 926 The general plan is to compile the code as a series of IF statements: 927 928 CASE (push 0 on the immediate-mode parameter stack) 929 test1 OF ... ENDOF test1 OVER = IF DROP ... ELSE 930 test2 OF ... ENDOF test2 OVER = IF DROP ... ELSE 931 testn OF ... ENDOF testn OVER = IF DROP ... ELSE 932 ... ( default case ) ... 933 ENDCASE DROP THEN [THEN [THEN ...]] 934 935 The CASE statement pushes 0 on the immediate-mode parameter stack, and that number 936 is used to count how many THEN statements we need when we get to ENDCASE so that each 937 IF has a matching THEN. The counting is done implicitly. If you recall from the 938 implementation above of IF, each IF pushes a code address on the immediate-mode stack, 939 and these addresses are non-zero, so by the time we get to ENDCASE the stack contains 940 some number of non-zeroes, followed by a zero. The number of non-zeroes is how many 941 times IF has been called, so how many times we need to match it with THEN. 942 943 This code uses [COMPILE] so that we compile calls to IF, ELSE, THEN instead of 944 actually calling them while we're compiling the words below. 945 946 As is the case with all of our control structures, they only work within word 947 definitions, not in immediate mode. 948 ) 949 : CASE IMMEDIATE 950 0 ( push 0 to mark the bottom of the stack ) 951 ; 952 953 : OF IMMEDIATE 954 ' OVER , ( compile OVER ) 955 ' = , ( compile = ) 956 [COMPILE] IF ( compile IF ) 957 ' DROP , ( compile DROP ) 958 ; 959 960 : ENDOF IMMEDIATE 961 [COMPILE] ELSE ( ENDOF is the same as ELSE ) 962 ; 963 964 : ENDCASE IMMEDIATE 965 ' DROP , ( compile DROP ) 966 967 ( keep compiling THEN until we get to our zero marker ) 968 BEGIN 969 ?DUP 970 WHILE 971 [COMPILE] THEN 972 REPEAT 973 ; 974 975 ( 976 DECOMPILER ---------------------------------------------------------------------- 977 978 CFA> is the opposite of >CFA. It takes a codeword and tries to find the matching 979 dictionary definition. (In truth, it works with any pointer into a word, not just 980 the codeword pointer, and this is needed to do stack traces). 981 982 In this FORTH this is not so easy. In fact we have to search through the dictionary 983 because we don't have a convenient back-pointer (as is often the case in other versions 984 of FORTH). Because of this search, CFA> should not be used when performance is critical, 985 so it is only used for debugging tools such as the decompiler and printing stack 986 traces. 987 988 This word returns 0 if it doesn't find a match. 989 ) 990 : CFA> 991 LATEST @ ( start at LATEST dictionary entry ) 992 BEGIN 993 ?DUP ( while link pointer is not null ) 994 WHILE 995 2DUP SWAP ( cfa curr curr cfa ) 996 < IF ( current dictionary entry < cfa? ) 997 NIP ( leave curr dictionary entry on the stack ) 998 EXIT 999 THEN 1000 @ ( follow link pointer back ) 1001 REPEAT 1002 DROP ( restore stack ) 1003 0 ( sorry, nothing found ) 1004 ; 1005 1006 ( 1007 SEE decompiles a FORTH word. 1008 1009 We search for the dictionary entry of the word, then search again for the next 1010 word (effectively, the end of the compiled word). This results in two pointers: 1011 1012 +---------+---+---+---+---+------------+------------+------------+------------+ 1013 | LINK | 3 | T | E | N | DOCOL | LIT | 10 | EXIT | 1014 +---------+---+---+---+---+------------+------------+------------+------------+ 1015 ^ ^ 1016 | | 1017 Start of word End of word 1018 1019 With this information we can have a go at decompiling the word. We need to 1020 recognise "meta-words" like LIT, LITSTRING, BRANCH, etc. and treat those separately. 1021 ) 1022 : SEE 1023 WORD FIND ( find the dictionary entry to decompile ) 1024 1025 ( Now we search again, looking for the next word in the dictionary. This gives us 1026 the length of the word that we will be decompiling. (Well, mostly it does). ) 1027 HERE @ ( address of the end of the last compiled word ) 1028 LATEST @ ( word last curr ) 1029 BEGIN 1030 2 PICK ( word last curr word ) 1031 OVER ( word last curr word curr ) 1032 <> ( word last curr word<>curr? ) 1033 WHILE ( word last curr ) 1034 NIP ( word curr ) 1035 DUP @ ( word curr prev (which becomes: word last curr) ) 1036 REPEAT 1037 1038 DROP ( at this point, the stack is: start-of-word end-of-word ) 1039 SWAP ( end-of-word start-of-word ) 1040 1041 ( begin the definition with : NAME [IMMEDIATE] ) 1042 ':' EMIT SPACE DUP ID. SPACE 1043 DUP ?IMMEDIATE IF ." IMMEDIATE " THEN 1044 1045 >DFA ( get the data address, ie. points after DOCOL | end-of-word start-of-data ) 1046 1047 ( now we start decompiling until we hit the end of the word ) 1048 BEGIN ( end start ) 1049 2DUP > 1050 WHILE 1051 DUP @ ( end start codeword ) 1052 1053 CASE 1054 ' LIT OF ( is it LIT ? ) 1055 4 + DUP @ ( get next word which is the integer constant ) 1056 . ( and print it ) 1057 ENDOF 1058 ' LITSTRING OF ( is it LITSTRING ? ) 1059 [ CHAR S ] LITERAL EMIT '"' EMIT SPACE ( print S"<space> ) 1060 4 + DUP @ ( get the length word ) 1061 SWAP 4 + SWAP ( end start+4 length ) 1062 2DUP TELL ( print the string ) 1063 '"' EMIT SPACE ( finish the string with a final quote ) 1064 + ALIGNED ( end start+4+len, aligned ) 1065 4 - ( because we're about to add 4 below ) 1066 ENDOF 1067 ' 0BRANCH OF ( is it 0BRANCH ? ) 1068 ." 0BRANCH ( " 1069 4 + DUP @ ( print the offset ) 1070 . 1071 ." ) " 1072 ENDOF 1073 ' BRANCH OF ( is it BRANCH ? ) 1074 ." BRANCH ( " 1075 4 + DUP @ ( print the offset ) 1076 . 1077 ." ) " 1078 ENDOF 1079 ' ' OF ( is it ' (TICK) ? ) 1080 [ CHAR ' ] LITERAL EMIT SPACE 1081 4 + DUP @ ( get the next codeword ) 1082 CFA> ( and force it to be printed as a dictionary entry ) 1083 ID. SPACE 1084 ENDOF 1085 ' EXIT OF ( is it EXIT? ) 1086 ( We expect the last word to be EXIT, and if it is then we don't print it 1087 because EXIT is normally implied by ;. EXIT can also appear in the middle 1088 of words, and then it needs to be printed. ) 1089 2DUP ( end start end start ) 1090 4 + ( end start end start+4 ) 1091 <> IF ( end start | we're not at the end ) 1092 ." EXIT " 1093 THEN 1094 ENDOF 1095 ( default case: ) 1096 DUP ( in the default case we always need to DUP before using ) 1097 CFA> ( look up the codeword to get the dictionary entry ) 1098 ID. SPACE ( and print it ) 1099 ENDCASE 1100 1101 4 + ( end start+4 ) 1102 REPEAT 1103 1104 ';' EMIT CR 1105 1106 2DROP ( restore stack ) 1107 ; 1108 1109 ( 1110 EXECUTION TOKENS ---------------------------------------------------------------------- 1111 1112 Standard FORTH defines a concept called an 'execution token' (or 'xt') which is very 1113 similar to a function pointer in C. We map the execution token to a codeword address. 1114 1115 execution token of DOUBLE is the address of this codeword 1116 | 1117 V 1118 +---------+---+---+---+---+---+---+---+---+------------+------------+------------+------------+ 1119 | LINK | 6 | D | O | U | B | L | E | 0 | DOCOL | DUP | + | EXIT | 1120 +---------+---+---+---+---+---+---+---+---+------------+------------+------------+------------+ 1121 len pad codeword ^ 1122 1123 There is one assembler primitive for execution tokens, EXECUTE ( xt -- ), which runs them. 1124 1125 You can make an execution token for an existing word the long way using >CFA, 1126 ie: WORD [foo] FIND >CFA will push the xt for foo onto the stack where foo is the 1127 next word in input. So a very slow way to run DOUBLE might be: 1128 1129 : DOUBLE DUP + ; 1130 : SLOW WORD FIND >CFA EXECUTE ; 1131 5 SLOW DOUBLE . CR \ prints 10 1132 1133 We also offer a simpler and faster way to get the execution token of any word FOO: 1134 1135 ['] FOO 1136 1137 (Exercises for readers: (1) What is the difference between ['] FOO and ' FOO? 1138 (2) What is the relationship between ', ['] and LIT?) 1139 1140 More useful is to define anonymous words and/or to assign xt's to variables. 1141 1142 To define an anonymous word (and push its xt on the stack) use :NONAME ... ; as in this 1143 example: 1144 1145 :NONAME ." anon word was called" CR ; \ pushes xt on the stack 1146 DUP EXECUTE EXECUTE \ executes the anon word twice 1147 1148 Stack parameters work as expected: 1149 1150 :NONAME ." called with parameter " . CR ; 1151 DUP 1152 10 SWAP EXECUTE \ prints 'called with parameter 10' 1153 20 SWAP EXECUTE \ prints 'called with parameter 20' 1154 1155 Notice that the above code has a memory leak: the anonymous word is still compiled 1156 into the data segment, so even if you lose track of the xt, the word continues to 1157 occupy memory. A good way to keep track of the xt and thus avoid the memory leak is 1158 to assign it to a CONSTANT, VARIABLE or VALUE: 1159 1160 0 VALUE ANON 1161 :NONAME ." anon word was called" CR ; TO ANON 1162 ANON EXECUTE 1163 ANON EXECUTE 1164 1165 Another use of :NONAME is to create an array of functions which can be called quickly 1166 (think: fast switch statement). This example is adapted from the ANS FORTH standard: 1167 1168 10 CELLS ALLOT CONSTANT CMD-TABLE 1169 : SET-CMD CELLS CMD-TABLE + ! ; 1170 : CALL-CMD CELLS CMD-TABLE + @ EXECUTE ; 1171 1172 :NONAME ." alternate 0 was called" CR ; 0 SET-CMD 1173 :NONAME ." alternate 1 was called" CR ; 1 SET-CMD 1174 \ etc... 1175 :NONAME ." alternate 9 was called" CR ; 9 SET-CMD 1176 1177 0 CALL-CMD 1178 1 CALL-CMD 1179 ) 1180 1181 : :NONAME 1182 0 0 CREATE ( create a word with no name - we need a dictionary header because ; expects it ) 1183 HERE @ ( current HERE value is the address of the codeword, ie. the xt ) 1184 DOCOL , ( compile DOCOL (the codeword) ) 1185 ] ( go into compile mode ) 1186 ; 1187 1188 : ['] IMMEDIATE 1189 ' LIT , ( compile LIT ) 1190 ; 1191 1192 ( 1193 EXCEPTIONS ---------------------------------------------------------------------- 1194 1195 Amazingly enough, exceptions can be implemented directly in FORTH, in fact rather easily. 1196 1197 The general usage is as follows: 1198 1199 : FOO ( n -- ) THROW ; 1200 1201 : TEST-EXCEPTIONS 1202 25 ['] FOO CATCH \ execute 25 FOO, catching any exception 1203 ?DUP IF 1204 ." called FOO and it threw exception number: " 1205 . CR 1206 DROP \ we have to drop the argument of FOO (25) 1207 THEN 1208 ; 1209 \ prints: called FOO and it threw exception number: 25 1210 1211 CATCH runs an execution token and detects whether it throws any exception or not. The 1212 stack signature of CATCH is rather complicated: 1213 1214 ( a_n-1 ... a_1 a_0 xt -- r_m-1 ... r_1 r_0 0 ) if xt did NOT throw an exception 1215 ( a_n-1 ... a_1 a_0 xt -- ?_n-1 ... ?_1 ?_0 e ) if xt DID throw exception 'e' 1216 1217 where a_i and r_i are the (arbitrary number of) argument and return stack contents 1218 before and after xt is EXECUTEd. Notice in particular the case where an exception 1219 is thrown, the stack pointer is restored so that there are n of _something_ on the 1220 stack in the positions where the arguments a_i used to be. We don't really guarantee 1221 what is on the stack -- perhaps the original arguments, and perhaps other nonsense -- 1222 it largely depends on the implementation of the word that was executed. 1223 1224 THROW, ABORT and a few others throw exceptions. 1225 1226 Exception numbers are non-zero integers. By convention the positive numbers can be used 1227 for app-specific exceptions and the negative numbers have certain meanings defined in 1228 the ANS FORTH standard. (For example, -1 is the exception thrown by ABORT). 1229 1230 0 THROW does nothing. This is the stack signature of THROW: 1231 1232 ( 0 -- ) 1233 ( * e -- ?_n-1 ... ?_1 ?_0 e ) the stack is restored to the state from the corresponding CATCH 1234 1235 The implementation hangs on the definitions of CATCH and THROW and the state shared 1236 between them. 1237 1238 Up to this point, the return stack has consisted merely of a list of return addresses, 1239 with the top of the return stack being the return address where we will resume executing 1240 when the current word EXITs. However CATCH will push a more complicated 'exception stack 1241 frame' on the return stack. The exception stack frame records some things about the 1242 state of execution at the time that CATCH was called. 1243 1244 When called, THROW walks up the return stack (the process is called 'unwinding') until 1245 it finds the exception stack frame. It then uses the data in the exception stack frame 1246 to restore the state allowing execution to continue after the matching CATCH. (If it 1247 unwinds the stack and doesn't find the exception stack frame then it prints a message 1248 and drops back to the prompt, which is also normal behaviour for so-called 'uncaught 1249 exceptions'). 1250 1251 This is what the exception stack frame looks like. (As is conventional, the return stack 1252 is shown growing downwards from higher to lower memory addresses). 1253 1254 +------------------------------+ 1255 | return address from CATCH | Notice this is already on the 1256 | | return stack when CATCH is called. 1257 +------------------------------+ 1258 | original parameter stack | 1259 | pointer | 1260 +------------------------------+ ^ 1261 | exception stack marker | | 1262 | (EXCEPTION-MARKER) | | Direction of stack 1263 +------------------------------+ | unwinding by THROW. 1264 | 1265 | 1266 1267 The EXCEPTION-MARKER marks the entry as being an exception stack frame rather than an 1268 ordinary return address, and it is this which THROW "notices" as it is unwinding the 1269 stack. (If you want to implement more advanced exceptions such as TRY...WITH then 1270 you'll need to use a different value of marker if you want the old and new exception stack 1271 frame layouts to coexist). 1272 1273 What happens if the executed word doesn't throw an exception? It will eventually 1274 return and call EXCEPTION-MARKER, so EXCEPTION-MARKER had better do something sensible 1275 without us needing to modify EXIT. This nicely gives us a suitable definition of 1276 EXCEPTION-MARKER, namely a function that just drops the stack frame and itself 1277 returns (thus "returning" from the original CATCH). 1278 1279 One thing to take from this is that exceptions are a relatively lightweight mechanism 1280 in FORTH. 1281 ) 1282 1283 : EXCEPTION-MARKER 1284 RDROP ( drop the original parameter stack pointer ) 1285 0 ( there was no exception, this is the normal return path ) 1286 ; 1287 1288 : CATCH ( xt -- exn? ) 1289 DSP@ 4+ >R ( save parameter stack pointer (+4 because of xt) on the return stack ) 1290 ' EXCEPTION-MARKER 4+ ( push the address of the RDROP inside EXCEPTION-MARKER ... ) 1291 >R ( ... on to the return stack so it acts like a return address ) 1292 EXECUTE ( execute the nested function ) 1293 ; 1294 1295 : THROW ( n -- ) 1296 ?DUP IF ( only act if the exception code <> 0 ) 1297 RSP@ ( get return stack pointer ) 1298 BEGIN 1299 DUP R0 4- < ( RSP < R0 ) 1300 WHILE 1301 DUP @ ( get the return stack entry ) 1302 ' EXCEPTION-MARKER 4+ = IF ( found the EXCEPTION-MARKER on the return stack ) 1303 4+ ( skip the EXCEPTION-MARKER on the return stack ) 1304 RSP! ( restore the return stack pointer ) 1305 1306 ( Restore the parameter stack. ) 1307 DUP DUP DUP ( reserve some working space so the stack for this word 1308 doesn't coincide with the part of the stack being restored ) 1309 R> ( get the saved parameter stack pointer | n dsp ) 1310 4- ( reserve space on the stack to store n ) 1311 SWAP OVER ( dsp n dsp ) 1312 ! ( write n on the stack ) 1313 DSP! EXIT ( restore the parameter stack pointer, immediately exit ) 1314 THEN 1315 4+ 1316 REPEAT 1317 1318 ( No matching catch - print a message and restart the INTERPRETer. ) 1319 DROP 1320 1321 CASE 1322 0 1- OF ( ABORT ) 1323 ." ABORTED" CR 1324 ENDOF 1325 ( default case ) 1326 ." UNCAUGHT THROW " 1327 DUP . CR 1328 ENDCASE 1329 QUIT 1330 THEN 1331 ; 1332 1333 : ABORT ( -- ) 1334 0 1- THROW 1335 ; 1336 1337 ( Print a stack trace by walking up the return stack. ) 1338 : PRINT-STACK-TRACE 1339 RSP@ ( start at caller of this function ) 1340 BEGIN 1341 DUP R0 4- < ( RSP < R0 ) 1342 WHILE 1343 DUP @ ( get the return stack entry ) 1344 CASE 1345 ' EXCEPTION-MARKER 4+ OF ( is it the exception stack frame? ) 1346 ." CATCH ( DSP=" 1347 4+ DUP @ U. ( print saved stack pointer ) 1348 ." ) " 1349 ENDOF 1350 ( default case ) 1351 DUP 1352 CFA> ( look up the codeword to get the dictionary entry ) 1353 ?DUP IF ( and print it ) 1354 2DUP ( dea addr dea ) 1355 ID. ( print word from dictionary entry ) 1356 [ CHAR + ] LITERAL EMIT 1357 SWAP >DFA 4+ - . ( print offset ) 1358 THEN 1359 ENDCASE 1360 4+ ( move up the stack ) 1361 REPEAT 1362 DROP 1363 CR 1364 ; 1365 1366 ( 1367 C STRINGS ---------------------------------------------------------------------- 1368 1369 FORTH strings are represented by a start address and length kept on the stack or in memory. 1370 1371 Most FORTHs don't handle C strings, but we need them in order to access the process arguments 1372 and environment left on the stack by the Linux kernel, and to make some system calls. 1373 1374 Operation Input Output FORTH word Notes 1375 ---------------------------------------------------------------------- 1376 1377 Create FORTH string addr len S" ..." 1378 1379 Create C string c-addr Z" ..." 1380 1381 C -> FORTH c-addr addr len DUP STRLEN 1382 1383 FORTH -> C addr len c-addr CSTRING Allocated in a temporary buffer, so 1384 should be consumed / copied immediately. 1385 FORTH string should not contain NULs. 1386 1387 For example, DUP STRLEN TELL prints a C string. 1388 ) 1389 1390 ( 1391 Z" .." is like S" ..." except that the string is terminated by an ASCII NUL character. 1392 1393 To make it more like a C string, at runtime Z" just leaves the address of the string 1394 on the stack (not address & length as with S"). To implement this we need to add the 1395 extra NUL to the string and also a DROP instruction afterwards. Apart from that the 1396 implementation just a modified S". 1397 ) 1398 : Z" IMMEDIATE 1399 STATE @ IF ( compiling? ) 1400 ' LITSTRING , ( compile LITSTRING ) 1401 HERE @ ( save the address of the length word on the stack ) 1402 0 , ( dummy length - we don't know what it is yet ) 1403 BEGIN 1404 KEY ( get next character of the string ) 1405 DUP '"' <> 1406 WHILE 1407 HERE @ C! ( store the character in the compiled image ) 1408 1 HERE +! ( increment HERE pointer by 1 byte ) 1409 REPEAT 1410 0 HERE @ C! ( add the ASCII NUL byte ) 1411 1 HERE +! 1412 DROP ( drop the double quote character at the end ) 1413 DUP ( get the saved address of the length word ) 1414 HERE @ SWAP - ( calculate the length ) 1415 4- ( subtract 4 (because we measured from the start of the length word) ) 1416 SWAP ! ( and back-fill the length location ) 1417 ALIGN ( round up to next multiple of 4 bytes for the remaining code ) 1418 ' DROP , ( compile DROP (to drop the length) ) 1419 ELSE ( immediate mode ) 1420 HERE @ ( get the start address of the temporary space ) 1421 BEGIN 1422 KEY 1423 DUP '"' <> 1424 WHILE 1425 OVER C! ( save next character ) 1426 1+ ( increment address ) 1427 REPEAT 1428 DROP ( drop the final " character ) 1429 0 SWAP C! ( store final ASCII NUL ) 1430 HERE @ ( push the start address ) 1431 THEN 1432 ; 1433 1434 : STRLEN ( str -- len ) 1435 DUP ( save start address ) 1436 BEGIN 1437 DUP C@ 0<> ( zero byte found? ) 1438 WHILE 1439 1+ 1440 REPEAT 1441 1442 SWAP - ( calculate the length ) 1443 ; 1444 1445 : CSTRING ( addr len -- c-addr ) 1446 SWAP OVER ( len saddr len ) 1447 HERE @ SWAP ( len saddr daddr len ) 1448 CMOVE ( len ) 1449 1450 HERE @ + ( daddr+len ) 1451 0 SWAP C! ( store terminating NUL char ) 1452 1453 HERE @ ( push start address ) 1454 ; 1455 1456 ( 1457 THE ENVIRONMENT ---------------------------------------------------------------------- 1458 1459 Linux makes the process arguments and environment available to us on the stack. 1460 1461 The top of stack pointer is saved by the early assembler code when we start up in the FORTH 1462 variable S0, and starting at this pointer we can read out the command line arguments and the 1463 environment. 1464 1465 Starting at S0, S0 itself points to argc (the number of command line arguments). 1466 1467 S0+4 points to argv[0], S0+8 points to argv[1] etc up to argv[argc-1]. 1468 1469 argv[argc] is a NULL pointer. 1470 1471 After that the stack contains environment variables, a set of pointers to strings of the 1472 form NAME=VALUE and on until we get to another NULL pointer. 1473 1474 The first word that we define, ARGC, pushes the number of command line arguments (note that 1475 as with C argc, this includes the name of the command). 1476 ) 1477 : ARGC 1478 S0 @ @ 1479 ; 1480 1481 ( 1482 n ARGV gets the nth command line argument. 1483 1484 For example to print the command name you would do: 1485 0 ARGV TELL CR 1486 ) 1487 : ARGV ( n -- str u ) 1488 1+ CELLS S0 @ + ( get the address of argv[n] entry ) 1489 @ ( get the address of the string ) 1490 DUP STRLEN ( and get its length / turn it into a FORTH string ) 1491 ; 1492 1493 ( 1494 ENVIRON returns the address of the first environment string. The list of strings ends 1495 with a NULL pointer. 1496 1497 For example to print the first string in the environment you could do: 1498 ENVIRON @ DUP STRLEN TELL 1499 ) 1500 : ENVIRON ( -- addr ) 1501 ARGC ( number of command line parameters on the stack to skip ) 1502 2 + ( skip command line count and NULL pointer after the command line args ) 1503 CELLS ( convert to an offset ) 1504 S0 @ + ( add to base stack address ) 1505 ; 1506 1507 ( 1508 SYSTEM CALLS AND FILES ---------------------------------------------------------------------- 1509 1510 Miscellaneous words related to system calls, and standard access to files. 1511 ) 1512 1513 ( BYE exits by calling the Linux exit(2) syscall. ) 1514 : BYE ( -- ) 1515 0 ( return code (0) ) 1516 SYS_EXIT ( system call number ) 1517 SYSCALL1 1518 ; 1519 1520 ( 1521 UNUSED returns the number of cells remaining in the user memory (data segment). 1522 1523 For our implementation we will use Linux brk(2) system call to find out the end 1524 of the data segment and subtract HERE from it. 1525 ) 1526 : GET-BRK ( -- brkpoint ) 1527 0 SYS_BRK SYSCALL1 ( call brk(0) ) 1528 ; 1529 1530 : UNUSED ( -- n ) 1531 GET-BRK ( get end of data segment according to the kernel ) 1532 HERE @ ( get current position in data segment ) 1533 - 1534 4 / ( returns number of cells ) 1535 ; 1536 1537 ( 1538 MORECORE increases the data segment by the specified number of (4 byte) cells. 1539 1540 NB. The number of cells requested should normally be a multiple of 1024. The 1541 reason is that Linux can't extend the data segment by less than a single page 1542 (4096 bytes or 1024 cells). 1543 1544 This FORTH doesn't automatically increase the size of the data segment "on demand" 1545 (ie. when , (COMMA), ALLOT, CREATE, and so on are used). Instead the programmer 1546 needs to be aware of how much space a large allocation will take, check UNUSED, and 1547 call MORECORE if necessary. A simple programming exercise is to change the 1548 implementation of the data segment so that MORECORE is called automatically if 1549 the program needs more memory. 1550 ) 1551 : BRK ( brkpoint -- ) 1552 SYS_BRK SYSCALL1 1553 ; 1554 1555 : MORECORE ( cells -- ) 1556 CELLS GET-BRK + BRK 1557 ; 1558 1559 ( 1560 Standard FORTH provides some simple file access primitives which we model on 1561 top of Linux syscalls. 1562 1563 The main complication is converting FORTH strings (address & length) into C 1564 strings for the Linux kernel. 1565 1566 Notice there is no buffering in this implementation. 1567 ) 1568 1569 : R/O ( -- fam ) O_RDONLY ; 1570 : R/W ( -- fam ) O_RDWR ; 1571 1572 : OPEN-FILE ( addr u fam -- fd 0 (if successful) | c-addr u fam -- fd errno (if there was an error) ) 1573 -ROT ( fam addr u ) 1574 CSTRING ( fam cstring ) 1575 SYS_OPEN SYSCALL2 ( open (filename, flags) ) 1576 DUP ( fd fd ) 1577 DUP 0< IF ( errno? ) 1578 NEGATE ( fd errno ) 1579 ELSE 1580 DROP 0 ( fd 0 ) 1581 THEN 1582 ; 1583 1584 : CREATE-FILE ( addr u fam -- fd 0 (if successful) | c-addr u fam -- fd errno (if there was an error) ) 1585 O_CREAT OR 1586 O_TRUNC OR 1587 -ROT ( fam addr u ) 1588 CSTRING ( fam cstring ) 1589 420 -ROT ( 0644 fam cstring ) 1590 SYS_OPEN SYSCALL3 ( open (filename, flags|O_TRUNC|O_CREAT, 0644) ) 1591 DUP ( fd fd ) 1592 DUP 0< IF ( errno? ) 1593 NEGATE ( fd errno ) 1594 ELSE 1595 DROP 0 ( fd 0 ) 1596 THEN 1597 ; 1598 1599 : CLOSE-FILE ( fd -- 0 (if successful) | fd -- errno (if there was an error) ) 1600 SYS_CLOSE SYSCALL1 1601 NEGATE 1602 ; 1603 1604 : READ-FILE ( addr u fd -- u2 0 (if successful) | addr u fd -- 0 0 (if EOF) | addr u fd -- u2 errno (if error) ) 1605 >R SWAP R> ( u addr fd ) 1606 SYS_READ SYSCALL3 1607 1608 DUP ( u2 u2 ) 1609 DUP 0< IF ( errno? ) 1610 NEGATE ( u2 errno ) 1611 ELSE 1612 DROP 0 ( u2 0 ) 1613 THEN 1614 ; 1615 1616 ( 1617 PERROR prints a message for an errno, similar to C's perror(3) but we don't have the extensive 1618 list of strerror strings available, so all we can do is print the errno. 1619 ) 1620 : PERROR ( errno addr u -- ) 1621 TELL 1622 ':' EMIT SPACE 1623 ." ERRNO=" 1624 . CR 1625 ; 1626 1627 ( 1628 ASSEMBLER CODE ---------------------------------------------------------------------- 1629 1630 This is just the outline of a simple assembler, allowing you to write FORTH primitives 1631 in assembly language. 1632 1633 Assembly primitives begin ': NAME' in the normal way, but are ended with ;CODE. ;CODE 1634 updates the header so that the codeword isn't DOCOL, but points instead to the assembled 1635 code (in the DFA part of the word). 1636 1637 We provide a convenience macro NEXT (you guessed what it does). However you don't need to 1638 use it because ;CODE will put a NEXT at the end of your word. 1639 1640 The rest consists of some immediate words which expand into machine code appended to the 1641 definition of the word. Only a very tiny part of the i386 assembly space is covered, just 1642 enough to write a few assembler primitives below. 1643 ) 1644 1645 HEX 1646 1647 ( Equivalent to the NEXT macro ) 1648 : NEXT IMMEDIATE AD C, FF C, 20 C, ; 1649 1650 : ;CODE IMMEDIATE 1651 [COMPILE] NEXT ( end the word with NEXT macro ) 1652 ALIGN ( machine code is assembled in bytes so isn't necessarily aligned at the end ) 1653 LATEST @ DUP 1654 HIDDEN ( unhide the word ) 1655 DUP >DFA SWAP >CFA ! ( change the codeword to point to the data area ) 1656 [COMPILE] [ ( go back to immediate mode ) 1657 ; 1658 1659 ( The i386 registers ) 1660 : EAX IMMEDIATE 0 ; 1661 : ECX IMMEDIATE 1 ; 1662 : EDX IMMEDIATE 2 ; 1663 : EBX IMMEDIATE 3 ; 1664 : ESP IMMEDIATE 4 ; 1665 : EBP IMMEDIATE 5 ; 1666 : ESI IMMEDIATE 6 ; 1667 : EDI IMMEDIATE 7 ; 1668 1669 ( i386 stack instructions ) 1670 : PUSH IMMEDIATE 50 + C, ; 1671 : POP IMMEDIATE 58 + C, ; 1672 1673 ( RDTSC instruction ) 1674 : RDTSC IMMEDIATE 0F C, 31 C, ; 1675 1676 DECIMAL 1677 1678 ( 1679 RDTSC is an assembler primitive which reads the Pentium timestamp counter (a very fine- 1680 grained counter which counts processor clock cycles). Because the TSC is 64 bits wide 1681 we have to push it onto the stack in two slots. 1682 ) 1683 : RDTSC ( -- lsb msb ) 1684 RDTSC ( writes the result in %edx:%eax ) 1685 EAX PUSH ( push lsb ) 1686 EDX PUSH ( push msb ) 1687 ;CODE 1688 1689 ( 1690 INLINE can be used to inline an assembler primitive into the current (assembler) 1691 word. 1692 1693 For example: 1694 1695 : 2DROP INLINE DROP INLINE DROP ;CODE 1696 1697 will build an efficient assembler word 2DROP which contains the inline assembly code 1698 for DROP followed by DROP (eg. two 'pop %eax' instructions in this case). 1699 1700 Another example. Consider this ordinary FORTH definition: 1701 1702 : C@++ ( addr -- addr+1 byte ) DUP 1+ SWAP C@ ; 1703 1704 (it is equivalent to the C operation '*p++' where p is a pointer to char). If we 1705 notice that all of the words used to define C@++ are in fact assembler primitives, 1706 then we can write a faster (but equivalent) definition like this: 1707 1708 : C@++ INLINE DUP INLINE 1+ INLINE SWAP INLINE C@ ;CODE 1709 1710 One interesting point to note is that this "concatenative" style of programming 1711 allows you to write assembler words portably. The above definition would work 1712 for any CPU architecture. 1713 1714 There are several conditions that must be met for INLINE to be used successfully: 1715 1716 (1) You must be currently defining an assembler word (ie. : ... ;CODE). 1717 1718 (2) The word that you are inlining must be known to be an assembler word. If you try 1719 to inline a FORTH word, you'll get an error message. 1720 1721 (3) The assembler primitive must be position-independent code and must end with a 1722 single NEXT macro. 1723 1724 Exercises for the reader: (a) Generalise INLINE so that it can inline FORTH words when 1725 building FORTH words. (b) Further generalise INLINE so that it does something sensible 1726 when you try to inline FORTH into assembler and vice versa. 1727 1728 The implementation of INLINE is pretty simple. We find the word in the dictionary, 1729 check it's an assembler word, then copy it into the current definition, byte by byte, 1730 until we reach the NEXT macro (which is not copied). 1731 ) 1732 HEX 1733 : =NEXT ( addr -- next? ) 1734 DUP C@ AD <> IF DROP FALSE EXIT THEN 1735 1+ DUP C@ FF <> IF DROP FALSE EXIT THEN 1736 1+ C@ 20 <> IF FALSE EXIT THEN 1737 TRUE 1738 ; 1739 DECIMAL 1740 1741 ( (INLINE) is the lowlevel inline function. ) 1742 : (INLINE) ( cfa -- ) 1743 @ ( remember codeword points to the code ) 1744 BEGIN ( copy bytes until we hit NEXT macro ) 1745 DUP =NEXT NOT 1746 WHILE 1747 DUP C@ C, 1748 1+ 1749 REPEAT 1750 DROP 1751 ; 1752 1753 : INLINE IMMEDIATE 1754 WORD FIND ( find the word in the dictionary ) 1755 >CFA ( codeword ) 1756 1757 DUP @ DOCOL = IF ( check codeword <> DOCOL (ie. not a FORTH word) ) 1758 ." Cannot INLINE FORTH words" CR ABORT 1759 THEN 1760 1761 (INLINE) 1762 ; 1763 1764 HIDE =NEXT 1765 1766 ( 1767 NOTES ---------------------------------------------------------------------- 1768 1769 DOES> isn't possible to implement with this FORTH because we don't have a separate 1770 data pointer. 1771 ) 1772 1773 ( 1774 WELCOME MESSAGE ---------------------------------------------------------------------- 1775 1776 Print the version and OK prompt. 1777 ) 1778 1779 : WELCOME 1780 S" TEST-MODE" FIND NOT IF 1781 ." JONESFORTH VERSION " VERSION . CR 1782 UNUSED . ." CELLS REMAINING" CR 1783 ." OK " 1784 THEN 1785 ; 1786 1787 WELCOME 1788 HIDE WELCOME 1789 1790