File:  [gforth] / gforth / prims2x.fs
Revision 1.122: download - view: text, annotated - select for diffs
Sun Dec 8 09:57:53 2002 UTC (21 years, 3 months ago) by anton
Branches: MAIN
CVS tags: HEAD
now engine/prim_num.i is generated, and used in main.c (for main-native.o)

    1: \ converts primitives to, e.g., C code 
    2: 
    3: \ Copyright (C) 1995,1996,1997,1998,2000 Free Software Foundation, Inc.
    4: 
    5: \ This file is part of Gforth.
    6: 
    7: \ Gforth is free software; you can redistribute it and/or
    8: \ modify it under the terms of the GNU General Public License
    9: \ as published by the Free Software Foundation; either version 2
   10: \ of the License, or (at your option) any later version.
   11: 
   12: \ This program is distributed in the hope that it will be useful,
   13: \ but WITHOUT ANY WARRANTY; without even the implied warranty of
   14: \ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
   15: \ GNU General Public License for more details.
   16: 
   17: \ You should have received a copy of the GNU General Public License
   18: \ along with this program; if not, write to the Free Software
   19: \ Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111, USA.
   20: 
   21: 
   22: \ This is not very nice (hard limits, no checking, assumes 1 chars = 1).
   23: \ And it grew even worse when it aged.
   24: 
   25: \ Optimizations:
   26: \ superfluous stores are removed. GCC removes the superfluous loads by itself
   27: \ TOS and FTOS can be kept in register( variable)s.
   28: \ 
   29: \ Problems:
   30: \ The TOS optimization is somewhat hairy. The problems by example:
   31: \ 1) dup ( w -- w w ): w=TOS; sp-=1; sp[1]=w; TOS=w;
   32: \    The store is not superfluous although the earlier opt. would think so
   33: \    Alternatively:    sp[0]=TOS; w=TOS; sp-=1; TOS=w;
   34: \ 2) ( -- .. ): sp[0] = TOS; ... /* This additional store is necessary */
   35: \ 3) ( .. -- ): ... TOS = sp[0]; /* as well as this load */
   36: \ 4) ( -- ): /* but here they are unnecessary */
   37: \ 5) Words that call NEXT themselves have to be done very carefully.
   38: \
   39: \ To do:
   40: \ add the store optimization for doubles
   41: \ regarding problem 1 above: It would be better (for over) to implement
   42: \ 	the alternative
   43: \ store optimization for combined instructions.
   44: 
   45: \ Design Uglyness:
   46: 
   47: \ - global state (values, variables) in connection with combined instructions.
   48: 
   49: \ - index computation is different for instruction-stream and the
   50: \ stacks; there are two mechanisms for dealing with that
   51: \ (stack-in-index-xt and a test for stack==instruction-stream); there
   52: \ should be only one.
   53: 
   54: warnings off
   55: 
   56: [IFUNDEF] try
   57: include startup.fs
   58: [THEN]
   59: 
   60: : struct% struct ; \ struct is redefined in gray
   61: 
   62: warnings off
   63: \ warnings on
   64: 
   65: include ./gray.fs
   66: 32 constant max-effect \ number of things on one side of a stack effect
   67: 4 constant max-stacks  \ the max. number of stacks (including inst-stream).
   68: 255 constant maxchar
   69: maxchar 1+ constant eof-char
   70: #tab constant tab-char
   71: #lf constant nl-char
   72: 
   73: variable rawinput \ pointer to next character to be scanned
   74: variable endrawinput \ pointer to the end of the input (the char after the last)
   75: variable cookedinput \ pointer to the next char to be parsed
   76: variable line \ line number of char pointed to by input
   77: variable line-start \ pointer to start of current line (for error messages)
   78: 0 line !
   79: 2variable filename \ filename of original input file
   80: 0 0 filename 2!
   81: 2variable out-filename \ filename of the output file (for sync lines)
   82: 0 0 out-filename 2!
   83: 2variable f-comment
   84: 0 0 f-comment 2!
   85: variable skipsynclines \ are sync lines ("#line ...") invisible to the parser?
   86: skipsynclines on
   87: variable out-nls \ newlines in output (for output sync lines)
   88: 0 out-nls !
   89: variable store-optimization \ use store optimization?
   90: store-optimization off
   91: 
   92: variable include-skipped-insts
   93: \ does the threaded code for a combined instruction include the cells
   94: \ for the component instructions (true) or only the cells for the
   95: \ inline arguments (false)
   96: include-skipped-insts off
   97: 
   98: variable immarg \ values for immediate arguments (to be used in IMM_ARG macros)
   99: $12340000 immarg !
  100: 
  101: : th ( addr1 n -- addr2 )
  102:     cells + ;
  103: 
  104: : holds ( addr u -- )
  105:     \ like HOLD, but for a string
  106:     tuck + swap 0 +do
  107: 	1- dup c@ hold
  108:     loop
  109:     drop ;
  110: 
  111: : insert-wordlist { c-addr u wordlist xt -- }
  112:     \ adds name "addr u" to wordlist using defining word xt
  113:     \ xt may cause additional stack effects
  114:     get-current >r wordlist set-current
  115:     c-addr u nextname xt execute
  116:     r> set-current ;
  117: 
  118: : start ( -- addr )
  119:  cookedinput @ ;
  120: 
  121: : end ( addr -- addr u )
  122:  cookedinput @ over - ;
  123: 
  124: : print-error-line ( -- )
  125:     \ print the current line and position
  126:     line-start @ endrawinput @ over - 2dup nl-char scan drop nip ( start end )
  127:     over - type cr
  128:     line-start @ rawinput @ over - typewhite ." ^" cr ;
  129: 
  130: : ?print-error { f addr u -- }
  131:     f ?not? if
  132: 	outfile-id >r try
  133: 	    stderr to outfile-id
  134: 	    filename 2@ type ." :" line @ 0 .r ." : " addr u type cr
  135: 	    print-error-line
  136: 	    0
  137: 	recover endtry
  138: 	r> to outfile-id throw
  139: 	1 (bye) \ abort
  140:     endif ;
  141: 
  142: : quote ( -- )
  143:     [char] " emit ;
  144: 
  145: \ count output lines to generate sync lines for output
  146: 
  147: : count-nls ( addr u -- )
  148:     bounds u+do
  149: 	i c@ nl-char = negate out-nls +!
  150:     loop ;
  151: 
  152: :noname ( addr u -- )
  153:     2dup count-nls
  154:     defers type ;
  155: is type
  156: 
  157: variable output          \ xt ( -- ) of output word for simple primitives
  158: variable output-combined \ xt ( -- ) of output word for combined primitives
  159: 
  160: struct%
  161:     cell%    field stack-number \ the number of this stack
  162:     cell% 2* field stack-pointer \ stackpointer name
  163:     cell%    field stack-type \ name for default type of stack items
  164:     cell%    field stack-in-index-xt \ ( in-size item -- in-index )
  165: end-struct stack%
  166: 
  167: struct%
  168:  cell% 2* field item-name   \ name, excluding stack prefixes
  169:  cell%    field item-stack  \ descriptor for the stack used, 0 is default
  170:  cell%    field item-type   \ descriptor for the item type
  171:  cell%    field item-offset \ offset in stack items, 0 for the deepest element
  172:  cell%	  field item-first  \ true if this is the first occurence of the item
  173: end-struct item%
  174: 
  175: struct%
  176:     cell% 2* field type-c-name
  177:     cell%    field type-stack \ default stack
  178:     cell%    field type-size  \ size of type in stack items
  179:     cell%    field type-fetch \ xt of fetch code generator ( item -- )
  180:     cell%    field type-store \ xt of store code generator ( item -- )
  181: end-struct type%
  182: 
  183: variable next-stack-number 0 next-stack-number !
  184: create stacks max-stacks cells allot \ array of stacks
  185: 
  186: : stack-in-index ( in-size item -- in-index )
  187:     item-offset @ - 1- ;
  188: 
  189: : inst-in-index ( in-size item -- in-index )
  190:     nip dup item-offset @ swap item-type @ type-size @ + 1- ;
  191: 
  192: : make-stack ( addr-ptr u1 type "stack-name" -- )
  193:     next-stack-number @ max-stacks < s" too many stacks" ?print-error
  194:     create stack% %allot >r
  195:     r@ stacks next-stack-number @ th !
  196:     next-stack-number @ r@ stack-number !
  197:     1 next-stack-number +!
  198:     r@ stack-type !
  199:     save-mem r@ stack-pointer 2! 
  200:     ['] stack-in-index r> stack-in-index-xt ! ;
  201: 
  202: : map-stacks { xt -- }
  203:     \ perform xt for all stacks
  204:     next-stack-number @ 0 +do
  205: 	stacks i th @ xt execute
  206:     loop ;
  207: 
  208: : map-stacks1 { xt -- }
  209:     \ perform xt for all stacks except inst-stream
  210:     next-stack-number @ 1 +do
  211: 	stacks i th @ xt execute
  212:     loop ;
  213: 
  214: \ stack items
  215: 
  216: : init-item ( addr u addr1 -- )
  217:     \ initialize item at addr1 with name addr u
  218:     \ !! remove stack prefix
  219:     dup item% %size erase
  220:     item-name 2! ;
  221: 
  222: : map-items { addr end xt -- }
  223:     \ perform xt for all items in array addr...end
  224:     end addr ?do
  225: 	i xt execute
  226:     item% %size +loop ;
  227: 
  228: \ types
  229: 
  230: : print-type-prefix ( type -- )
  231:     body> >head name>string type ;
  232: 
  233: \ various variables for storing stuff of one primitive
  234: 
  235: struct%
  236:     cell% 2* field prim-name
  237:     cell% 2* field prim-wordset
  238:     cell% 2* field prim-c-name
  239:     cell% 2* field prim-doc
  240:     cell% 2* field prim-c-code
  241:     cell% 2* field prim-forth-code
  242:     cell% 2* field prim-stack-string
  243:     cell%    field prim-num            \ ordinal number
  244:     cell%    field prim-items-wordlist \ unique items
  245:     item% max-effect * field prim-effect-in
  246:     item% max-effect * field prim-effect-out
  247:     cell%    field prim-effect-in-end
  248:     cell%    field prim-effect-out-end
  249:     cell% max-stacks * field prim-stacks-in  \ number of in items per stack
  250:     cell% max-stacks * field prim-stacks-out \ number of out items per stack
  251: end-struct prim%
  252: 
  253: : make-prim ( -- prim )
  254:     prim% %alloc { p }
  255:     s" " p prim-doc 2! s" " p prim-forth-code 2! s" " p prim-wordset 2!
  256:     p ;
  257: 
  258: 0 value prim     \ in combined prims either combined or a part
  259: 0 value combined \ in combined prims the combined prim
  260: variable in-part \ true if processing a part
  261:  in-part off
  262: 
  263: : prim-context ( ... p xt -- ... )
  264:     \ execute xt with prim set to p
  265:     prim >r
  266:     swap to prim
  267:     catch
  268:     r> to prim
  269:     throw ;
  270: 
  271: 1000 constant max-combined
  272: create combined-prims max-combined cells allot
  273: variable num-combined
  274: variable part-num \ current part number during process-combined
  275: 
  276: : map-combined { xt -- }
  277:     \ perform xt for all components of the current combined instruction
  278:     num-combined @ 0 +do
  279: 	combined-prims i th @ xt execute
  280:     loop ;
  281: 
  282: table constant combinations
  283:   \ the keys are the sequences of pointers to primitives
  284: 
  285: create current-depth max-stacks cells allot
  286: create max-depth     max-stacks cells allot
  287: create min-depth     max-stacks cells allot
  288: 
  289: create sp-update-in max-stacks cells allot
  290: \ where max-depth occured the first time
  291: create max-depths max-stacks max-combined 1+ * cells allot
  292: \ maximum depth at start of each part: array[parts] of array[stack]
  293: create max-back-depths max-stacks max-combined 1+ * cells allot
  294: \ maximun depth from end of the combination to the start of the each part
  295: 
  296: : s-c-max-depth ( nstack ncomponent -- addr )
  297:     max-stacks * + cells max-depths + ;
  298: 
  299: : s-c-max-back-depth ( nstack ncomponent -- addr )
  300:     max-stacks * + cells max-back-depths + ;
  301: 
  302: wordlist constant primitives
  303: 
  304: : create-prim ( prim -- )
  305:     dup prim-name 2@ primitives ['] constant insert-wordlist ;
  306: 
  307: : stack-in ( stack -- addr )
  308:     \ address of number of stack items in effect in
  309:     stack-number @ cells prim prim-stacks-in + ;
  310: 
  311: : stack-out ( stack -- addr )
  312:     \ address of number of stack items in effect out
  313:     stack-number @ cells prim prim-stacks-out + ;
  314: 
  315: \ global vars
  316: variable c-line
  317: 2variable c-filename
  318: variable name-line
  319: 2variable name-filename
  320: 2variable last-name-filename
  321: Variable function-number 0 function-number !
  322: 
  323: \ a few more set ops
  324: 
  325: : bit-equivalent ( w1 w2 -- w3 )
  326:  xor invert ;
  327: 
  328: : complement ( set1 -- set2 )
  329:  empty ['] bit-equivalent binary-set-operation ;
  330: 
  331: \ forward declaration for inst-stream (breaks cycle in definitions)
  332: defer inst-stream-f ( -- stack )
  333: 
  334: \ stack access stuff
  335: 
  336: : normal-stack-access1 ( n stack -- )
  337:     stack-pointer 2@ type
  338:     dup
  339:     if
  340: 	." [" 0 .r ." ]"
  341:     else
  342: 	drop ." TOS"
  343:     endif ;
  344: 
  345: : normal-stack-access ( n stack -- )
  346:     dup inst-stream-f = if
  347: 	." IMM_ARG(" normal-stack-access1 ." ," immarg ? ." )"
  348: 	1 immarg +!
  349:     else
  350: 	normal-stack-access1
  351:     endif ;
  352: 
  353: : stack-depth { stack -- n }
  354:     current-depth stack stack-number @ th @ ;
  355: 
  356: : part-stack-access { n stack -- }
  357:     \ print _<stack><x>, x=inst-stream? n : maxdepth-currentdepth-n-1
  358:     ." _" stack stack-pointer 2@ type
  359:     stack stack-number @ { stack# }
  360:     stack stack-depth n + { access-depth }
  361:     stack inst-stream-f = if
  362: 	access-depth
  363:     else
  364: 	combined prim-stacks-in stack# th @
  365: 	assert( dup max-depth stack# th @ = )
  366: 	access-depth - 1-
  367:     endif
  368:     0 .r ;
  369: 
  370: : part-stack-read { n stack -- }
  371:     stack stack-depth n + ( ndepth )
  372:     stack stack-number @ part-num @ s-c-max-depth @
  373: \    max-depth stack stack-number @ th @ ( ndepth nmaxdepth )
  374:     over <= if ( ndepth ) \ load from memory
  375: 	stack normal-stack-access
  376:     else
  377: 	drop n stack part-stack-access
  378:     endif ;
  379: 
  380: : stack-diff ( stack -- n )
  381:     \ in-out
  382:     dup stack-in @ swap stack-out @ - ;
  383: 
  384: : part-stack-write { n stack -- }
  385:     stack stack-depth n +
  386:     stack stack-number @ part-num @ s-c-max-back-depth @
  387:     over <= if ( ndepth )
  388: 	stack combined ['] stack-diff prim-context -
  389: 	stack normal-stack-access
  390:     else
  391: 	drop n stack part-stack-access
  392:     endif ;
  393: 
  394: : stack-read ( n stack -- )
  395:     \ print a stack access at index n of stack
  396:     in-part @ if
  397: 	part-stack-read
  398:     else
  399: 	normal-stack-access
  400:     endif ;
  401: 
  402: : stack-write ( n stack -- )
  403:     \ print a stack access at index n of stack
  404:     in-part @ if
  405: 	part-stack-write
  406:     else
  407: 	normal-stack-access
  408:     endif ;
  409: 
  410: : item-in-index { item -- n }
  411:     \ n is the index of item (in the in-effect)
  412:     item item-stack @ dup >r stack-in @ ( in-size r:stack )
  413:     item r> stack-in-index-xt @ execute ;
  414: 
  415: : item-stack-type-name ( item -- addr u )
  416:     item-stack @ stack-type @ type-c-name 2@ ;
  417: 
  418: : fetch-single ( item -- )
  419:     \ fetch a single stack item from its stack
  420:     >r
  421:     ." vm_" r@ item-stack-type-name type
  422:     ." 2" r@ item-type @ print-type-prefix ." ("
  423:     r@ item-in-index r@ item-stack @ stack-read ." ,"
  424:     r@ item-name 2@ type
  425:     ." );" cr
  426:     rdrop ; 
  427: 
  428: : fetch-double ( item -- )
  429:     \ fetch a double stack item from its stack
  430:     >r
  431:     ." vm_two"
  432:     r@ item-stack-type-name type ." 2"
  433:     r@ item-type @ print-type-prefix ." ("
  434:     r@ item-in-index r@ item-stack @ 2dup ." (Cell)" stack-read
  435:     ." , "                      -1 under+ ." (Cell)" stack-read
  436:     ." , " r@ item-name 2@ type
  437:     ." )" cr
  438:     rdrop ;
  439: 
  440: : same-as-in? ( item -- f )
  441:  \ f is true iff the offset and stack of item is the same as on input
  442:  >r
  443:  r@ item-first @ if
  444:      rdrop false exit
  445:  endif
  446:  r@ item-name 2@ prim prim-items-wordlist @ search-wordlist 0= abort" bug"
  447:  execute @
  448:  dup r@ =
  449:  if \ item first appeared in output
  450:    drop false
  451:  else
  452:    dup  item-stack  @ r@ item-stack  @ = 
  453:    swap item-offset @ r@ item-offset @ = and
  454:  endif
  455:  rdrop ;
  456: 
  457: : item-out-index ( item -- n )
  458:     \ n is the index of item (in the in-effect)
  459:     >r r@ item-stack @ stack-out @ r> item-offset @ - 1- ;
  460: 
  461: : really-store-single ( item -- )
  462:     >r
  463:     ." vm_"
  464:     r@ item-type @ print-type-prefix ." 2"
  465:     r@ item-stack-type-name type ." ("
  466:     r@ item-name 2@ type ." ,"
  467:     r@ item-out-index r@ item-stack @ stack-write ." );"
  468:     rdrop ;
  469: 
  470: : store-single ( item -- )
  471:     >r
  472:     store-optimization @ in-part @ 0= and r@ same-as-in? and if
  473: 	r@ item-in-index 0= r@ item-out-index 0= xor if
  474: 	    ." IF_" r@ item-stack @ stack-pointer 2@ type
  475: 	    ." TOS(" r@ really-store-single ." );" cr
  476: 	endif
  477:     else
  478: 	r@ really-store-single cr
  479:     endif
  480:     rdrop ;
  481: 
  482: : store-double ( item -- )
  483: \ !! store optimization is not performed, because it is not yet needed
  484:  >r
  485:  ." vm_"
  486:  r@ item-type @ print-type-prefix ." 2two"
  487:  r@ item-stack-type-name type ." ("
  488:  r@ item-name 2@ type ." , "
  489:  r@ item-out-index r@ item-stack @ 2dup stack-write
  490:  ." , "                       -1 under+ stack-write
  491:  ." )" cr
  492:  rdrop ;
  493: 
  494: : single ( -- xt1 xt2 n )
  495:     ['] fetch-single ['] store-single 1 ;
  496: 
  497: : double ( -- xt1 xt2 n )
  498:     ['] fetch-double ['] store-double 2 ;
  499: 
  500: : s, ( addr u -- )
  501: \ allocate a string
  502:  here swap dup allot move ;
  503: 
  504: wordlist constant prefixes
  505: 
  506: : declare ( addr "name" -- )
  507: \ remember that there is a stack item at addr called name
  508:  create , ;
  509: 
  510: : !default ( w addr -- )
  511:     dup @ if
  512: 	2drop \ leave nonzero alone
  513:     else
  514: 	!
  515:     endif ;
  516: 
  517: : create-type { addr u xt1 xt2 n stack -- } ( "prefix" -- )
  518:     \ describes a type
  519:     \ addr u specifies the C type name
  520:     \ stack effect entries of the type start with prefix
  521:     create type% %allot >r
  522:     addr u save-mem r@ type-c-name 2!
  523:     xt1   r@ type-fetch !
  524:     xt2   r@ type-store !
  525:     n     r@ type-size !
  526:     stack r@ type-stack !
  527:     rdrop ;
  528: 
  529: : type-prefix ( addr u xt1 xt2 n stack "prefix" -- )
  530:     get-current >r prefixes set-current
  531:     create-type r> set-current
  532: does> ( item -- )
  533:     \ initialize item
  534:     { item typ }
  535:     typ item item-type !
  536:     typ type-stack @ item item-stack !default
  537:     item item-name 2@ prim prim-items-wordlist @ search-wordlist 0= if
  538: 	item item-name 2@ nextname item declare
  539: 	item item-first on
  540: 	\ typ type-c-name 2@ type space type  ." ;" cr
  541:     else
  542: 	drop
  543: 	item item-first off
  544:     endif ;
  545: 
  546: : execute-prefix ( item addr1 u1 -- )
  547:     \ execute the word ( item -- ) associated with the longest prefix
  548:     \ of addr1 u1
  549:     0 swap ?do
  550: 	dup i prefixes search-wordlist
  551: 	if \ ok, we have the type ( item addr1 xt )
  552: 	    nip execute
  553: 	    UNLOOP EXIT
  554: 	endif
  555: 	-1 s+loop
  556:     \ we did not find a type, abort
  557:     false s" unknown prefix" ?print-error ;
  558: 
  559: : declaration ( item -- )
  560:     dup item-name 2@ execute-prefix ;
  561: 
  562: : declaration-list ( addr1 addr2 -- )
  563:     ['] declaration map-items ;
  564: 
  565: : declarations ( -- )
  566:  wordlist dup prim prim-items-wordlist ! set-current
  567:  prim prim-effect-in prim prim-effect-in-end @ declaration-list
  568:  prim prim-effect-out prim prim-effect-out-end @ declaration-list ;
  569: 
  570: : print-declaration { item -- }
  571:     item item-first @ if
  572: 	item item-type @ type-c-name 2@ type space
  573: 	item item-name 2@ type ." ;" cr
  574:     endif ;
  575: 
  576: : print-declarations ( -- )
  577:     prim prim-effect-in  prim prim-effect-in-end  @ ['] print-declaration map-items
  578:     prim prim-effect-out prim prim-effect-out-end @ ['] print-declaration map-items ;
  579:     
  580: : stack-prefix ( stack "prefix" -- )
  581:     get-current >r prefixes set-current
  582:     name tuck nextname create ( stack length ) 2,
  583:     r> set-current
  584: does> ( item -- )
  585:     2@ { item stack prefix-length }
  586:     item item-name 2@ prefix-length /string item item-name 2!
  587:     stack item item-stack !
  588:     item declaration ;
  589: 
  590: \ types pointed to by stacks for use in combined prims
  591: \ !! output-c-combined shouldn't use these names!
  592: : stack-type-name ( addr u "name" -- )
  593:     single 0 create-type ;
  594: 
  595: wordlist constant type-names \ this is here just to meet the requirement
  596:                     \ that a type be a word; it is never used for lookup
  597: 
  598: : stack ( "name" "stack-pointer" "type" -- )
  599:     \ define stack
  600:     name { d: stack-name }
  601:     name { d: stack-pointer }
  602:     name { d: stack-type }
  603:     get-current type-names set-current
  604:     stack-type 2dup nextname stack-type-name
  605:     set-current
  606:     stack-pointer lastxt >body stack-name nextname make-stack ;
  607: 
  608: stack inst-stream IP Cell
  609: ' inst-in-index inst-stream stack-in-index-xt !
  610: ' inst-stream <is> inst-stream-f
  611: \ !! initialize stack-in and stack-out
  612: 
  613: \ offset computation
  614: \ the leftmost (i.e. deepest) item has offset 0
  615: \ the rightmost item has the highest offset
  616: 
  617: : compute-offset { item xt -- }
  618:     \ xt specifies in/out; update stack-in/out and set item-offset
  619:     item item-type @ type-size @
  620:     item item-stack @ xt execute dup @ >r +!
  621:     r> item item-offset ! ;
  622: 
  623: : compute-offset-in ( addr1 addr2 -- )
  624:     ['] stack-in compute-offset ;
  625: 
  626: : compute-offset-out ( addr1 addr2 -- )
  627:     ['] stack-out compute-offset ;
  628: 
  629: : clear-stack ( stack -- )
  630:     dup stack-in off stack-out off ;
  631: 
  632: : compute-offsets ( -- )
  633:     ['] clear-stack map-stacks
  634:     prim prim-effect-in  prim prim-effect-in-end  @ ['] compute-offset-in  map-items
  635:     prim prim-effect-out prim prim-effect-out-end @ ['] compute-offset-out map-items
  636:     inst-stream stack-out @ 0= s" # can only be on the input side" ?print-error ;
  637: 
  638: : process-simple ( -- )
  639:     prim prim { W^ key } key cell
  640:     combinations ['] constant insert-wordlist
  641:     declarations compute-offsets
  642:     output @ execute ;
  643: 
  644: : flush-a-tos { stack -- }
  645:     stack stack-out @ 0<> stack stack-in @ 0= and
  646:     if
  647: 	." IF_" stack stack-pointer 2@ 2dup type ." TOS("
  648: 	2dup type ." [0] = " type ." TOS);" cr
  649:     endif ;
  650: 
  651: : flush-tos ( -- )
  652:     ['] flush-a-tos map-stacks1 ;
  653: 
  654: : fill-a-tos { stack -- }
  655:     stack stack-out @ 0= stack stack-in @ 0<> and
  656:     if
  657: 	." IF_" stack stack-pointer 2@ 2dup type ." TOS("
  658: 	2dup type ." TOS = " type ." [0]);" cr
  659:     endif ;
  660: 
  661: : fill-tos ( -- )
  662:     \ !! inst-stream for prefetching?
  663:     ['] fill-a-tos map-stacks1 ;
  664: 
  665: : fetch ( addr -- )
  666:     dup item-type @ type-fetch @ execute ;
  667: 
  668: : fetches ( -- )
  669:     prim prim-effect-in prim prim-effect-in-end @ ['] fetch map-items ;
  670: 
  671: : stack-pointer-update { stack -- }
  672:     \ stack grow downwards
  673:     stack stack-diff
  674:     ?dup-if \ this check is not necessary, gcc would do this for us
  675: 	stack inst-stream = if
  676: 	    ." INC_IP(" 0 .r ." );" cr
  677: 	else
  678: 	    stack stack-pointer 2@ type ."  += " 0 .r ." ;" cr
  679: 	endif
  680:     endif ;
  681: 
  682: : stack-pointer-updates ( -- )
  683:     ['] stack-pointer-update map-stacks ;
  684: 
  685: : store ( item -- )
  686: \ f is true if the item should be stored
  687: \ f is false if the store is probably not necessary
  688:  dup item-type @ type-store @ execute ;
  689: 
  690: : stores ( -- )
  691:     prim prim-effect-out prim prim-effect-out-end @ ['] store map-items ;
  692: 
  693: : print-debug-arg { item -- }
  694:     ." fputs(" quote space item item-name 2@ type ." =" quote ." , vm_out); "
  695:     ." printarg_" item item-type @ print-type-prefix
  696:     ." (" item item-name 2@ type ." );" cr ;
  697:     
  698: : print-debug-args ( -- )
  699:     ." #ifdef VM_DEBUG" cr
  700:     ." if (vm_debug) {" cr
  701:     prim prim-effect-in prim prim-effect-in-end @ ['] print-debug-arg map-items
  702: \    ." fputc('\n', vm_out);" cr
  703:     ." }" cr
  704:     ." #endif" cr ;
  705: 
  706: : print-debug-result { item -- }
  707:     item item-first @ if
  708: 	item print-debug-arg
  709:     endif ;
  710: 
  711: : print-debug-results ( -- )
  712:     cr
  713:     ." #ifdef VM_DEBUG" cr
  714:     ." if (vm_debug) {" cr
  715:     ." fputs(" quote ."  -- " quote ." , vm_out); "
  716:     prim prim-effect-out prim prim-effect-out-end @ ['] print-debug-result map-items
  717:     ." fputc('\n', vm_out);" cr
  718:     ." }" cr
  719:     ." #endif" cr ;
  720: 
  721: : output-super-end ( -- )
  722:     prim prim-c-code 2@ s" SET_IP" search if
  723: 	." SUPER_END;" cr
  724:     endif
  725:     2drop ;
  726: 
  727: : output-label2 ( -- )
  728:     ." LABEL2(" prim prim-c-name 2@ type ." )" cr
  729:     ." NEXT_P2;" cr ;
  730: 
  731: : output-c-tail1 { xt -- }
  732:     \ the final part of the generated C code, with xt printing LABEL2 or not.
  733:     output-super-end
  734:     print-debug-results
  735:     ." NEXT_P1;" cr
  736:     stores
  737:     fill-tos 
  738:     xt execute ;
  739: 
  740: : output-c-tail1-no-stores { xt -- }
  741:     \ the final part of the generated C code for combinations
  742:     output-super-end
  743:     ." NEXT_P1;" cr
  744:     fill-tos 
  745:     xt execute ;
  746: 
  747: : output-c-tail ( -- )
  748:     ['] noop output-c-tail1 ;
  749: 
  750: : output-c-tail2 ( -- )
  751:     ['] output-label2 output-c-tail1 ;
  752: 
  753: : output-c-tail-no-stores ( -- )
  754:     ['] noop output-c-tail1-no-stores ;
  755: 
  756: : output-c-tail2-no-stores ( -- )
  757:     ['] output-label2 output-c-tail1-no-stores ;
  758: 
  759: : type-c-code ( c-addr u xt -- )
  760:     \ like TYPE, but replaces "INST_TAIL;" with tail code produced by xt
  761:     { xt }
  762:     ." {" cr
  763:     ." #line " c-line @ . quote c-filename 2@ type quote cr
  764:     begin ( c-addr1 u1 )
  765: 	2dup s" INST_TAIL;" search
  766:     while ( c-addr1 u1 c-addr3 u3 )
  767: 	2dup 2>r drop nip over - type
  768: 	xt execute
  769: 	2r> 10 /string
  770: 	\ !! resync #line missing
  771:     repeat
  772:     2drop type
  773:     ." #line " out-nls @ 2 + . quote out-filename 2@ type quote cr
  774:     ." }" cr ;
  775: 
  776: : print-entry ( -- )
  777:     ." LABEL(" prim prim-c-name 2@ type ." )" ;
  778:     
  779: : output-c ( -- ) 
  780:     print-entry ."  /* " prim prim-name 2@ type ."  ( " prim prim-stack-string 2@ type ." ) */" cr
  781:     ." /* " prim prim-doc 2@ type ."  */" cr
  782:     ." NAME(" quote prim prim-name 2@ type quote ." )" cr \ debugging
  783:     ." {" cr
  784:     ." DEF_CA" cr
  785:     print-declarations
  786:     ." NEXT_P0;" cr
  787:     flush-tos
  788:     fetches
  789:     print-debug-args
  790:     stack-pointer-updates
  791:     prim prim-c-code 2@ ['] output-c-tail type-c-code
  792:     output-c-tail2
  793:     ." }" cr
  794:     cr
  795: ;
  796: 
  797: : disasm-arg { item -- }
  798:     item item-stack @ inst-stream = if
  799: 	." {" cr
  800: 	item print-declaration
  801: 	item fetch
  802: 	item print-debug-arg
  803: 	." }" cr
  804:     endif ;
  805: 
  806: : disasm-args ( -- )
  807:     prim prim-effect-in prim prim-effect-in-end @ ['] disasm-arg map-items ;
  808: 
  809: : output-disasm ( -- )
  810:     \ generate code for disassembling VM instructions
  811:     ." if (VM_IS_INST(*ip, " function-number @ 0 .r ." )) {" cr
  812:     ."   fputs(" quote prim prim-name 2@ type quote ." , vm_out);" cr
  813:     disasm-args
  814:     ."   ip += " inst-stream stack-in @ 1+ 0 .r ." ;" cr
  815:     ."   goto _endif_;" cr
  816:     ." }" cr ;
  817: 
  818: : output-profile ( -- )
  819:     \ generate code for postprocessing the VM block profile stuff
  820:     ." if (VM_IS_INST(*ip, " function-number @ 0 .r ." )) {" cr
  821:     ."   add_inst(b, " quote prim prim-name 2@ type quote ." );" cr
  822:     ."   ip += " inst-stream stack-in @ 1+ 0 .r ." ;" cr
  823:     prim prim-c-code 2@  s" SET_IP"    search nip nip
  824:     prim prim-c-code 2@  s" SUPER_END" search nip nip or if
  825: 	."   return;" cr
  826:     else
  827: 	."   goto _endif_;" cr
  828:     endif
  829:     ." }" cr ;
  830: 
  831: : output-profile-part ( p )
  832:     ."   add_inst(b, " quote
  833:     prim-name 2@ type
  834:     quote ." );" cr ;
  835:     
  836: : output-profile-combined ( -- )
  837:     \ generate code for postprocessing the VM block profile stuff
  838:     ." if (VM_IS_INST(*ip, " function-number @ 0 .r ." )) {" cr
  839:     ['] output-profile-part map-combined
  840:     ."   ip += " inst-stream stack-in @ 1+ 0 .r ." ;" cr
  841:     combined-prims num-combined @ 1- th @ prim-c-code 2@  s" SET_IP"    search nip nip
  842:     combined-prims num-combined @ 1- th @ prim-c-code 2@  s" SUPER_END" search nip nip or if
  843: 	."   return;" cr
  844:     else
  845: 	."   goto _endif_;" cr
  846:     endif
  847:     ." }" cr ;
  848: 
  849: : output-superend ( -- )
  850:     \ output flag specifying whether the current word ends a dynamic superinst
  851:     prim prim-c-code 2@  s" SET_IP"    search nip nip
  852:     prim prim-c-code 2@  s" SUPER_END" search nip nip or 0<>
  853:     prim prim-c-code 2@  s" SUPER_CONTINUE" search nip nip 0= and
  854:     negate 0 .r ." , /* " prim prim-name 2@ type ."  */" cr ;
  855: 
  856: : gen-arg-parm { item -- }
  857:     item item-stack @ inst-stream = if
  858: 	." , " item item-type @ type-c-name 2@ type space
  859: 	item item-name 2@ type
  860:     endif ;
  861: 
  862: : gen-args-parm ( -- )
  863:     prim prim-effect-in prim prim-effect-in-end @ ['] gen-arg-parm map-items ;
  864: 
  865: : gen-arg-gen { item -- }
  866:     item item-stack @ inst-stream = if
  867: 	."   genarg_" item item-type @ print-type-prefix
  868:         ." (ctp, " item item-name 2@ type ." );" cr
  869:     endif ;
  870: 
  871: : gen-args-gen ( -- )
  872:     prim prim-effect-in prim prim-effect-in-end @ ['] gen-arg-gen map-items ;
  873: 
  874: : output-gen ( -- )
  875:     \ generate C code for generating VM instructions
  876:     ." void gen_" prim prim-c-name 2@ type ." (Inst **ctp" gen-args-parm ." )" cr
  877:     ." {" cr
  878:     ."   gen_inst(ctp, vm_prim[" function-number @ 0 .r ." ]);" cr
  879:     gen-args-gen
  880:     ." }" cr ;
  881: 
  882: : stack-used? { stack -- f }
  883:     stack stack-in @ stack stack-out @ or 0<> ;
  884: 
  885: : output-funclabel ( -- )
  886:   ." &I_" prim prim-c-name 2@ type ." ," cr ;
  887: 
  888: : output-forthname ( -- )
  889:   '" emit prim prim-name 2@ type '" emit ." ," cr ;
  890: 
  891: \  : output-c-func ( -- )
  892: \  \ used for word libraries
  893: \      ." Cell * I_" prim prim-c-name 2@ type ." (Cell *SP, Cell **FP)      /* " prim prim-name 2@ type
  894: \      ."  ( " prim prim-stack-string 2@ type ."  ) */" cr
  895: \      ." /* " prim prim-doc 2@ type ."  */" cr
  896: \      ." NAME(" quote prim prim-name 2@ type quote ." )" cr
  897: \      \ debugging
  898: \      ." {" cr
  899: \      print-declarations
  900: \      \ !! don't know what to do about that
  901: \      inst-stream  stack-used? IF ." Cell *ip=IP;" cr THEN
  902: \      data-stack   stack-used? IF ." Cell *sp=SP;" cr THEN
  903: \      fp-stack     stack-used? IF ." Cell *fp=*FP;" cr THEN
  904: \      return-stack stack-used? IF ." Cell *rp=*RP;" cr THEN
  905: \      flush-tos
  906: \      fetches
  907: \      stack-pointer-updates
  908: \      fp-stack   stack-used? IF ." *FP=fp;" cr THEN
  909: \      ." {" cr
  910: \      ." #line " c-line @ . quote c-filename 2@ type quote cr
  911: \      prim prim-c-code 2@ type
  912: \      ." }" cr
  913: \      stores
  914: \      fill-tos
  915: \      ." return (sp);" cr
  916: \      ." }" cr
  917: \      cr ;
  918: 
  919: : output-label ( -- )  
  920:     ." INST_ADDR(" prim prim-c-name 2@ type ." )," cr ;
  921: 
  922: : output-alias ( -- ) 
  923:     ( primitive-number @ . ." alias " ) ." Primitive " prim prim-name 2@ type cr ;
  924: 
  925: : output-c-prim-num ( -- )
  926:     ." #define N_" prim prim-c-name 2@ type prim prim-num @ 8 + 4 .r cr ;
  927: 
  928: : output-forth ( -- )  
  929:     prim prim-forth-code @ 0=
  930:     IF    	\ output-alias
  931: 	\ this is bad for ec: an alias is compiled if tho word does not exist!
  932: 	\ JAW
  933:     ELSE  ." : " prim prim-name 2@ type ."   ( "
  934: 	prim prim-stack-string 2@ type ." )" cr
  935: 	prim prim-forth-code 2@ type cr
  936:     THEN ;
  937: 
  938: : output-tag-file ( -- )
  939:     name-filename 2@ last-name-filename 2@ compare if
  940: 	name-filename 2@ last-name-filename 2!
  941: 	#ff emit cr
  942: 	name-filename 2@ type
  943: 	." ,0" cr
  944:     endif ;
  945: 
  946: : output-tag ( -- )
  947:     output-tag-file
  948:     prim prim-name 2@ 1+ type
  949:     127 emit
  950:     space prim prim-name 2@ type space
  951:     1 emit
  952:     name-line @ 0 .r
  953:     ." ,0" cr ;
  954: 
  955: : output-vi-tag ( -- )
  956:     name-filename 2@ type #tab emit
  957:     prim prim-name 2@ type #tab emit
  958:     ." /^" prim prim-name 2@ type ."  *(/" cr ;
  959: 
  960: [IFDEF] documentation
  961: : register-doc ( -- )
  962:     prim prim-name 2@ documentation ['] create insert-wordlist
  963:     prim prim-name 2@ 2,
  964:     prim prim-stack-string 2@ condition-stack-effect 2,
  965:     prim prim-wordset 2@ 2,
  966:     prim prim-c-name 2@ condition-pronounciation 2,
  967:     prim prim-doc 2@ 2, ;
  968: [THEN]
  969: 
  970: 
  971: \ combining instructions
  972: 
  973: \ The input should look like this:
  974: 
  975: \ lit_+ = lit +
  976: 
  977: \ The output should look like this:
  978: 
  979: \  I_lit_+:
  980: \  {
  981: \  DEF_CA
  982: \  Cell _x_ip0;
  983: \  Cell _x_sp0;
  984: \  Cell _x_sp1;
  985: \  NEXT_P0;
  986: \  _x_ip0 = (Cell) IPTOS;
  987: \  _x_sp0 = (Cell) spTOS;
  988: \  INC_IP(1);
  989: \  /* sp += 0; */
  990: \  /* lit ( #w -- w ) */
  991: \  /*  */
  992: \  NAME("lit")
  993: \  {
  994: \  Cell w;
  995: \  w = (Cell) _x_ip0;
  996: \  #ifdef VM_DEBUG
  997: \  if (vm_debug) {
  998: \  fputs(" w=", vm_out); printarg_w (w);
  999: \  fputc('\n', vm_out);
 1000: \  }
 1001: \  #endif
 1002: \  {
 1003: \  #line 136 "./prim"
 1004: \  }
 1005: \  _x_sp1 = (Cell)w;
 1006: \  }
 1007: \  I_plus:	/* + ( n1 n2 -- n ) */
 1008: \  /*  */
 1009: \  NAME("+")
 1010: \  {
 1011: \  DEF_CA
 1012: \  Cell n1;
 1013: \  Cell n2;
 1014: \  Cell n;
 1015: \  NEXT_P0;
 1016: \  n1 = (Cell) _x_sp0;
 1017: \  n2 = (Cell) _x_sp1;
 1018: \  #ifdef VM_DEBUG
 1019: \  if (vm_debug) {
 1020: \  fputs(" n1=", vm_out); printarg_n (n1);
 1021: \  fputs(" n2=", vm_out); printarg_n (n2);
 1022: \  fputc('\n', vm_out);
 1023: \  }
 1024: \  #endif
 1025: \  {
 1026: \  #line 516 "./prim"
 1027: \  n = n1+n2;
 1028: \  }
 1029: \  _x_sp0 = (Cell)n;
 1030: \  }
 1031: \  NEXT_P1;
 1032: \  spTOS = (Cell)_x_sp0;
 1033: \  NEXT_P2;
 1034: 
 1035: : init-combined ( -- )
 1036:     prim to combined
 1037:     0 num-combined !
 1038:     current-depth max-stacks cells erase
 1039:     include-skipped-insts @ current-depth 0 th !
 1040:     max-depth     max-stacks cells erase
 1041:     min-depth     max-stacks cells erase
 1042:     prim prim-effect-in  prim prim-effect-in-end  !
 1043:     prim prim-effect-out prim prim-effect-out-end ! ;
 1044: 
 1045: : max! ( n addr -- )
 1046:     tuck @ max swap ! ;
 1047: 
 1048: : min! ( n addr -- )
 1049:     tuck @ min swap ! ;
 1050: 
 1051: : inst-stream-adjustment ( nstack -- n )
 1052:     \ number of stack items to add for each part
 1053:     0= include-skipped-insts @ and negate ;
 1054: 
 1055: : add-depths { p -- }
 1056:     \ combine stack effect of p with *-depths
 1057:     max-stacks 0 ?do
 1058: 	current-depth i th @
 1059: 	p prim-stacks-in  i th @ + i inst-stream-adjustment +
 1060: 	dup max-depth i th max!
 1061: 	p prim-stacks-out i th @ -
 1062: 	dup min-depth i th min!
 1063: 	current-depth i th !
 1064:     loop ;
 1065: 
 1066: : copy-maxdepths ( n -- )
 1067:     max-depth max-depths rot max-stacks * th max-stacks cells move ;
 1068: 
 1069: : add-prim ( addr u -- )
 1070:     \ add primitive given by "addr u" to combined-prims
 1071:     primitives search-wordlist s" unknown primitive" ?print-error
 1072:     execute { p }
 1073:     p combined-prims num-combined @ th !
 1074:     num-combined @ copy-maxdepths
 1075:     1 num-combined +!
 1076:     p add-depths
 1077:     num-combined @ copy-maxdepths ;
 1078: 
 1079: : compute-effects { q -- }
 1080:     \ compute the stack effects of q from the depths
 1081:     max-stacks 0 ?do
 1082: 	max-depth i th @ dup
 1083: 	q prim-stacks-in i th !
 1084: 	current-depth i th @ -
 1085: 	q prim-stacks-out i th !
 1086:     loop ;
 1087: 
 1088: : make-effect-items { stack# items effect-endp -- }
 1089:     \ effect-endp points to a pointer to the end of the current item-array
 1090:     \ and has to be updated
 1091:     stacks stack# th @ { stack }
 1092:     items 0 +do
 1093: 	effect-endp @ { item }
 1094: 	i 0 <# #s stack stack-pointer 2@ holds [char] _ hold #> save-mem
 1095: 	item item-name 2!
 1096: 	stack item item-stack !
 1097: 	stack stack-type @ item item-type !
 1098: 	i item item-offset !
 1099: 	item item-first on
 1100: 	item% %size effect-endp +!
 1101:     loop ;
 1102: 
 1103: : init-effects { q -- }
 1104:     \ initialize effects field for FETCHES and STORES
 1105:     max-stacks 0 ?do
 1106: 	i q prim-stacks-in  i th @ q prim-effect-in-end  make-effect-items
 1107: 	i q prim-stacks-out i th @ q prim-effect-out-end make-effect-items
 1108:     loop ;
 1109: 
 1110: : compute-stack-max-back-depths ( stack -- )
 1111:     stack-number @ { stack# }
 1112:     current-depth stack# th @ dup
 1113:     dup stack# num-combined @ s-c-max-back-depth !
 1114:     -1 num-combined @ 1- -do ( max-depth current-depth )
 1115: 	combined-prims i th @ { p }
 1116: 	p prim-stacks-out stack# th @ +
 1117: 	dup >r max r>
 1118: 	over stack# i s-c-max-back-depth !
 1119: 	p prim-stacks-in stack# th @ -
 1120: 	stack# inst-stream-adjustment -
 1121:     1 -loop
 1122:     assert( dup stack# inst-stream-adjustment negate = )
 1123:     assert( over max-depth stack# th @ = )
 1124:     2drop ;
 1125: 
 1126: : compute-max-back-depths ( -- )
 1127:     \ compute max-back-depths.
 1128:     \ assumes that current-depths is correct for the end of the combination
 1129:     ['] compute-stack-max-back-depths map-stacks ;
 1130: 
 1131: : process-combined ( -- )
 1132:     combined combined-prims num-combined @ cells
 1133:     combinations ['] constant insert-wordlist
 1134:     combined-prims num-combined @ 1- th ( last-part )
 1135:     @ prim-c-code 2@ prim prim-c-code 2! \ used by output-super-end
 1136:     prim compute-effects
 1137:     prim init-effects
 1138:     compute-max-back-depths
 1139:     output-combined perform ;
 1140: 
 1141: \ C output
 1142: 
 1143: : print-item { n stack -- }
 1144:     \ print nth stack item name
 1145:     stack stack-type @ type-c-name 2@ type space
 1146:     ." _" stack stack-pointer 2@ type n 0 .r ;
 1147: 
 1148: : print-declarations-combined ( -- )
 1149:     max-stacks 0 ?do
 1150: 	max-depth i th @ min-depth i th @ - 0 +do
 1151: 	    i stacks j th @ print-item ." ;" cr
 1152: 	loop
 1153:     loop ;
 1154: 
 1155: : part-fetches ( -- )
 1156:     fetches ;
 1157: 
 1158: : part-output-c-tail ( -- )
 1159:     print-debug-results
 1160:     stores ;
 1161: 
 1162: : output-combined-tail ( -- )
 1163:     part-output-c-tail
 1164:     in-part @ >r in-part off
 1165:     combined ['] output-c-tail-no-stores prim-context
 1166:     r> in-part ! ;
 1167: 
 1168: : part-stack-pointer-updates ( -- )
 1169:     max-stacks 0 +do
 1170: 	i part-num @ 1+ s-c-max-depth @ dup
 1171: 	i num-combined @ s-c-max-depth @ =    \ final depth
 1172: 	swap i part-num @ s-c-max-depth @ <> \ just reached now
 1173: 	part-num @ 0= \ first part
 1174: 	or and if
 1175: 	    stacks i th @ stack-pointer-update
 1176: 	endif
 1177:     loop ;
 1178: 
 1179: : output-part ( p -- )
 1180:     to prim
 1181:     ." /* " prim prim-name 2@ type ."  ( " prim prim-stack-string 2@ type ." ) */" cr
 1182:     ." NAME(" quote prim prim-name 2@ type quote ." )" cr \ debugging
 1183:     ." {" cr
 1184:     print-declarations
 1185:     part-fetches
 1186:     print-debug-args
 1187:     combined ['] part-stack-pointer-updates prim-context
 1188:     1 part-num +!
 1189:     prim add-depths \ !! right place?
 1190:     prim prim-c-code 2@ ['] output-combined-tail type-c-code
 1191:     part-output-c-tail
 1192:     ." }" cr ;
 1193: 
 1194: : output-parts ( -- )
 1195:     prim >r in-part on
 1196:     current-depth max-stacks cells erase
 1197:     0 part-num !
 1198:     ['] output-part map-combined
 1199:     in-part off
 1200:     r> to prim ;
 1201: 
 1202: : output-c-combined ( -- )
 1203:     print-entry cr
 1204:     \ debugging messages just in parts
 1205:     ." {" cr
 1206:     ." DEF_CA" cr
 1207:     print-declarations-combined
 1208:     ." NEXT_P0;" cr
 1209:     flush-tos
 1210:     \ fetches \ now in parts
 1211:     \ print-debug-args
 1212:     \ stack-pointer-updates now in parts
 1213:     output-parts
 1214:     output-c-tail2-no-stores
 1215:     ." }" cr
 1216:     cr ;
 1217: 
 1218: : output-forth-combined ( -- )
 1219: ;
 1220: 
 1221: 
 1222: \ peephole optimization rules
 1223: 
 1224: \ data for a simple peephole optimizer that always tries to combine
 1225: \ the currently compiled instruction with the last one.
 1226: 
 1227: \ in order for this to work as intended, shorter combinations for each
 1228: \ length must be present, and the longer combinations must follow
 1229: \ shorter ones (this restriction may go away in the future).
 1230:   
 1231: : output-peephole ( -- )
 1232:     combined-prims num-combined @ 1- cells combinations search-wordlist
 1233:     s" the prefix for this superinstruction must be defined earlier" ?print-error
 1234:     ." {"
 1235:     execute prim-num @ 5 .r ." ,"
 1236:     combined-prims num-combined @ 1- th @ prim-num @ 5 .r ." ,"
 1237:     combined prim-num @ 5 .r ." }, /* "
 1238:     combined prim-c-name 2@ type ."  */"
 1239:     cr ;
 1240: 
 1241: 
 1242: \ cost and superinstruction data for a sophisticated combiner (e.g.,
 1243: \ shortest path)
 1244: 
 1245: \ This is intended as initializer for a structure like this
 1246: 
 1247: \  struct cost {
 1248: \    int loads;       /* number of stack loads */
 1249: \    int stores;      /* number of stack stores */
 1250: \    int updates;     /* number of stack pointer updates */
 1251: \    int length;      /* number of components */
 1252: \    int *components; /* array of vm_prim indexes of components */
 1253: \  };
 1254: 
 1255: \ How do you know which primitive or combined instruction this
 1256: \ structure refers to?  By the order of cost structures, as in most
 1257: \ other cases.
 1258: 
 1259: : compute-costs { p -- nloads nstores nupdates }
 1260:     \ compute the number of loads, stores, and stack pointer updates
 1261:     \ of a primitive or combined instruction; does not take TOS
 1262:     \ caching into account, nor that IP updates are combined with
 1263:     \ other stuff
 1264:     0 max-stacks 0 +do
 1265: 	p prim-stacks-in i th @ +
 1266:     loop
 1267:     0 max-stacks 0 +do
 1268: 	p prim-stacks-out i th @ +
 1269:     loop
 1270:     0 max-stacks 0 +do
 1271: 	p prim-stacks-in i th @ p prim-stacks-out i th @ <> -
 1272:     loop ;
 1273: 
 1274: : output-num-part ( p -- )
 1275:     prim-num @ 4 .r ." ," ;
 1276: 
 1277: : output-costs ( -- )
 1278:     ." {" prim compute-costs
 1279:     rot 2 .r ." ," swap 2 .r ." ," 2 .r ." ,"
 1280:     combined if
 1281: 	num-combined @ 2 .r
 1282: 	." , ((int []){" ['] output-num-part map-combined ." })}, /* "
 1283:     else
 1284: 	."  1, ((int []){" prim prim-num @ 4 .r ." })}, /* "
 1285:     endif
 1286:     prim prim-name 2@ type ."  */"
 1287:     cr ;
 1288: 
 1289: \ the parser
 1290: 
 1291: eof-char max-member \ the whole character set + EOF
 1292: 
 1293: : getinput ( -- n )
 1294:  rawinput @ endrawinput @ =
 1295:  if
 1296:    eof-char
 1297:  else
 1298:    cookedinput @ c@
 1299:  endif ;
 1300: 
 1301: :noname ( n -- )
 1302:  dup bl > if
 1303:   emit space
 1304:  else
 1305:   .
 1306:  endif ;
 1307: print-token !
 1308: 
 1309: : testchar? ( set -- f )
 1310:  getinput member? ;
 1311: ' testchar? test-vector !
 1312: 
 1313: : checksyncline ( -- )
 1314:     \ when input points to a newline, check if the next line is a
 1315:     \ sync line.  If it is, perform the appropriate actions.
 1316:     rawinput @ >r
 1317:     s" #line " r@ over compare if
 1318: 	rdrop 1 line +! EXIT
 1319:     endif
 1320:     0. r> 6 chars + 20 >number drop >r drop line ! r> ( c-addr )
 1321:     dup c@ bl = if
 1322: 	char+ dup c@ [char] " <> 0= s" sync line syntax" ?print-error
 1323: 	char+ dup 100 [char] " scan drop swap 2dup - save-mem filename 2!
 1324: 	char+
 1325:     endif
 1326:     dup c@ nl-char <> 0= s" sync line syntax" ?print-error
 1327:     skipsynclines @ if
 1328: 	dup char+ rawinput !
 1329: 	rawinput @ c@ cookedinput @ c!
 1330:     endif
 1331:     drop ;
 1332: 
 1333: : ?nextchar ( f -- )
 1334:     s" syntax error, wrong char" ?print-error
 1335:     rawinput @ endrawinput @ <> if
 1336: 	rawinput @ c@
 1337: 	1 chars rawinput +!
 1338: 	1 chars cookedinput +!
 1339: 	nl-char = if
 1340: 	    checksyncline
 1341: 	    rawinput @ line-start !
 1342: 	endif
 1343: 	rawinput @ c@ cookedinput @ c!
 1344:     endif ;
 1345: 
 1346: : charclass ( set "name" -- )
 1347:  ['] ?nextchar terminal ;
 1348: 
 1349: : .. ( c1 c2 -- set )
 1350:  ( creates a set that includes the characters c, c1<=c<=c2 )
 1351:  empty copy-set
 1352:  swap 1+ rot do
 1353:   i over add-member
 1354:  loop ;
 1355: 
 1356: : ` ( -- terminal ) ( use: ` c )
 1357:  ( creates anonymous terminal for the character c )
 1358:  char singleton ['] ?nextchar make-terminal ;
 1359: 
 1360: char a char z ..  char A char Z ..  union char _ singleton union  charclass letter
 1361: char 0 char 9 ..					charclass digit
 1362: bl singleton tab-char over add-member			charclass white
 1363: nl-char singleton eof-char over add-member complement	charclass nonl
 1364: nl-char singleton eof-char over add-member
 1365:     char : over add-member complement                   charclass nocolonnl
 1366: nl-char singleton eof-char over add-member
 1367:     char } over add-member complement                   charclass nobracenl
 1368: bl 1+ maxchar .. char \ singleton complement intersection
 1369:                                                         charclass nowhitebq
 1370: bl 1+ maxchar ..                                        charclass nowhite
 1371: char " singleton eof-char over add-member complement	charclass noquote
 1372: nl-char singleton					charclass nl
 1373: eof-char singleton					charclass eof
 1374: nl-char singleton eof-char over add-member		charclass nleof
 1375: 
 1376: (( letter (( letter || digit )) **
 1377: )) <- c-ident ( -- )
 1378: 
 1379: (( ` # ?? (( letter || digit || ` : )) ++
 1380: )) <- stack-ident ( -- )
 1381: 
 1382: (( nowhitebq nowhite ** ))
 1383: <- forth-ident ( -- )
 1384: 
 1385: Variable forth-flag
 1386: Variable c-flag
 1387: 
 1388: (( (( ` e || ` E )) {{ start }} nonl ** 
 1389:    {{ end evaluate }}
 1390: )) <- eval-comment ( ... -- ... )
 1391: 
 1392: (( (( ` f || ` F )) {{ start }} nonl ** 
 1393:    {{ end forth-flag @ IF type cr ELSE 2drop THEN }}
 1394: )) <- forth-comment ( -- )
 1395: 
 1396: (( (( ` c || ` C )) {{ start }} nonl ** 
 1397:    {{ end c-flag @ IF type cr ELSE 2drop THEN }}
 1398: )) <- c-comment ( -- )
 1399: 
 1400: (( ` - nonl ** {{ 
 1401: 	forth-flag @ IF ." [ELSE]" cr THEN
 1402: 	c-flag @ IF ." #else" cr THEN }}
 1403: )) <- else-comment
 1404: 
 1405: (( ` + {{ start }} nonl ** {{ end
 1406: 	dup
 1407: 	IF	c-flag @
 1408: 		IF    ." #ifdef HAS_" bounds ?DO  I c@ toupper emit  LOOP cr
 1409: 		THEN
 1410: 		forth-flag @
 1411: 		IF  ." has? " type ."  [IF]"  cr THEN
 1412: 	ELSE	2drop
 1413: 	    c-flag @      IF  ." #endif"  cr THEN
 1414: 	    forth-flag @  IF  ." [THEN]"  cr THEN
 1415: 	THEN }}
 1416: )) <- if-comment
 1417: 
 1418: (( (( ` g || ` G )) {{ start }} nonl **
 1419:    {{ end
 1420:       forth-flag @ IF  ." group " type cr  THEN
 1421:       c-flag @     IF  ." GROUP(" type ." )" cr  THEN }}
 1422: )) <- group-comment
 1423: 
 1424: (( (( eval-comment || forth-comment || c-comment || else-comment || if-comment || group-comment )) ?? nonl ** )) <- comment-body
 1425: 
 1426: (( ` \ comment-body nleof )) <- comment ( -- )
 1427: 
 1428: (( {{ start }} stack-ident {{ end 2 pick init-item item% %size + }} white ** )) **
 1429: <- stack-items
 1430: 
 1431: (( {{ prim prim-effect-in }}  stack-items {{ prim prim-effect-in-end ! }}
 1432:    ` - ` - white **
 1433:    {{ prim prim-effect-out }} stack-items {{ prim prim-effect-out-end ! }}
 1434: )) <- stack-effect ( -- )
 1435: 
 1436: (( {{ prim create-prim }}
 1437:    ` ( white ** {{ start }} stack-effect {{ end prim prim-stack-string 2! }} ` ) white **
 1438:    (( {{ start }} forth-ident {{ end prim prim-wordset 2! }} white **
 1439:       (( {{ start }}  c-ident {{ end prim prim-c-name 2! }} )) ??
 1440:    )) ??  nleof
 1441:    (( ` " ` "  {{ start }} (( noquote ++ ` " )) ++ {{ end 1- prim prim-doc 2! }} ` " white ** nleof )) ??
 1442:    {{ skipsynclines off line @ c-line ! filename 2@ c-filename 2! start }}
 1443:    (( (( ` { nonl ** nleof (( (( nobracenl {{ line @ drop }} nonl ** )) ?? nleof )) ** ` } white ** nleof white ** ))
 1444:    || (( nocolonnl nonl **  nleof white ** )) ** ))
 1445:    {{ end prim prim-c-code 2! skipsynclines on }}
 1446:    (( ` :  white ** nleof
 1447:       {{ start }} (( nonl ++  nleof white ** )) ++ {{ end prim prim-forth-code 2! }}
 1448:    )) ?? {{ process-simple }}
 1449:    nleof
 1450: )) <- simple-primitive ( -- )
 1451: 
 1452: (( {{ init-combined }}
 1453:    ` = white ** (( {{ start }} forth-ident {{ end add-prim }} white ** )) ++
 1454:    nleof {{ process-combined }}
 1455: )) <- combined-primitive
 1456: 
 1457: (( {{ make-prim to prim 0 to combined
 1458:       line @ name-line ! filename 2@ name-filename 2!
 1459:       function-number @ prim prim-num !
 1460:       start }} [ifdef] vmgen c-ident [else] forth-ident [then] {{ end
 1461:       2dup prim prim-name 2! prim prim-c-name 2! }}  white **
 1462:    (( ` / white ** {{ start }} c-ident {{ end prim prim-c-name 2! }} white ** )) ??
 1463:    (( simple-primitive || combined-primitive )) {{ 1 function-number +! }}
 1464: )) <- primitive ( -- )
 1465: 
 1466: (( (( comment || primitive || nl white ** )) ** eof ))
 1467: parser primitives2something
 1468: warnings @ [IF]
 1469: .( parser generated ok ) cr
 1470: [THEN]
 1471: 
 1472: 
 1473: \ run with gforth-0.5.0 (slurp-file is missing)
 1474: [IFUNDEF] slurp-file
 1475: : slurp-file ( c-addr1 u1 -- c-addr2 u2 )
 1476:     \ c-addr1 u1 is the filename, c-addr2 u2 is the file's contents
 1477:     r/o bin open-file throw >r
 1478:     r@ file-size throw abort" file too large"
 1479:     dup allocate throw swap
 1480:     2dup r@ read-file throw over <> abort" could not read whole file"
 1481:     r> close-file throw ;
 1482: [THEN]
 1483: 
 1484: : primfilter ( addr u -- )
 1485:     \ process the string at addr u
 1486:     over dup rawinput ! dup line-start ! cookedinput !
 1487:     + endrawinput !
 1488:     checksyncline
 1489:     primitives2something ;    
 1490: 
 1491: : process-file ( addr u xt-simple x-combined -- )
 1492:     output-combined ! output !
 1493:     save-mem 2dup filename 2!
 1494:     slurp-file
 1495:     warnings @ if
 1496: 	." ------------ CUT HERE -------------" cr  endif
 1497:     primfilter ;
 1498: 
 1499: \  : process      ( xt -- )
 1500: \      bl word count rot
 1501: \      process-file ;

FreeBSD-CVSweb <freebsd-cvsweb@FreeBSD.org>