| blob | 9a86abe1dff752c5b060fafd323a6f9d4993b8a5 |
1 ;;;; the top level interfaces to the compiler, plus some other
2 ;;;; compiler-related stuff (e.g. CL:CALL-ARGUMENTS-LIMIT) which
3 ;;;; doesn't obviously belong anywhere else
5 ;;;; This software is part of the SBCL system. See the README file for
6 ;;;; more information.
7 ;;;;
8 ;;;; This software is derived from the CMU CL system, which was
9 ;;;; written at Carnegie Mellon University and released into the
10 ;;;; public domain. The software is in the public domain and is
11 ;;;; provided with absolutely no warranty. See the COPYING and CREDITS
12 ;;;; files for more information.
16 ;;; FIXME: Doesn't this belong somewhere else, like early-c.lisp?
34 ;;; Whether call of a function which cannot be defined causes a full
35 ;;; warning.
41 ;;; Set to NIL to disable loop analysis for register allocation.
44 ;;; Bind this to a stream to capture various internal debugging output.
47 ;;; The current block compilation state. These are initialized to the
48 ;;; :BLOCK-COMPILE and :ENTRY-POINTS arguments that COMPILE-FILE was
49 ;;; called with.
50 ;;;
51 ;;; *BLOCK-COMPILE-ARG* holds the original value of the :BLOCK-COMPILE
52 ;;; argument, which overrides any internal declarations.
59 ;;; When block compiling, used by PROCESS-FORM to accumulate top level
60 ;;; lambdas resulting from compiling subforms. (In reverse order.)
65 #!+sb-doc
68 #!+sb-doc
71 #!+sb-doc
72 "When this is true, the compiler prints to *ERROR-OUTPUT* progress
73 information about the phases of compilation of each function. (This
77 #!+sb-doc
78 "The defaulted pathname of the file currently being compiled, or NIL if not
81 #!+sb-doc
82 "The TRUENAME of the file currently being compiled, or NIL if not
89 ;;; the SOURCE-INFO structure for the current compilation. This is
90 ;;; null globally to indicate that we aren't currently in any
91 ;;; identifiable compilation.
94 ;;; This is true if we are within a WITH-COMPILATION-UNIT form (which
95 ;;; normally causes nested uses to be no-ops).
98 ;;; Count of the number of compilation units dynamically enclosed by
99 ;;; the current active WITH-COMPILATION-UNIT that were unwound out of.
102 ;;; Mumble conditional on *COMPILE-PROGRESS*.
113 \f
114 ;;;; WITH-COMPILATION-UNIT and WITH-COMPILATION-VALUES
117 #!+sb-doc
118 "WITH-COMPILATION-UNIT ({Key Value}*) Form*
119 This form affects compilations that take place within its dynamic extent. It
120 is intended to be wrapped around the compilation of all files in the same
121 system. These keywords are defined:
122 :OVERRIDE Boolean-Form
123 One of the effects of this form is to delay undefined warnings
124 until the end of the form, instead of giving them at the end of each
125 compilation. If OVERRIDE is NIL (the default), then the outermost
126 WITH-COMPILATION-UNIT form grabs the undefined warnings. Specifying
127 OVERRIDE true causes that form to grab any enclosed warnings, even if
128 it is enclosed by another WITH-COMPILATION-UNIT."
135 ;; Inside another WITH-COMPILATION-UNIT, a WITH-COMPILATION-UNIT is
136 ;; ordinarily (unless OVERRIDE) basically a no-op.
160 ;;; Is FUN-NAME something that no conforming program can rely on
161 ;;; defining as a function?
166 ;;; This is to be called at the end of a compilation unit. It signals
167 ;;; any residual warnings about unknown stuff, then prints the total
168 ;;; error counts. ABORT-P should be true when the compilation unit was
169 ;;; aborted by throwing out. ABORT-COUNT is the number of dynamically
170 ;;; enclosed nested compilation units that were aborted.
191 #+sb-xc-host nil
195 "~@<There is no function named ~S. References to ~S in ~
196 some contexts (like starts of blocks) have special ~
197 meaning, but here it would have to be a function, ~
198 and that shouldn't be right.~:@>"
199 name name))
202 "~@<The ~(~A~) ~S is undefined, and its name is ~
203 reserved by ANSI CL so that even if it were ~
204 defined later, the code doing so would not be ~
205 portable.~:@>"
206 kind name)))
215 "~W more use~:P of undefined ~(~A~) ~S"
216 more kind name)
218 "~W more use~:P of undefined ~(~A~) ~S"
219 more kind name)))))))
228 "~:[This ~(~A~) is~;These ~(~A~)s are~] undefined:~
229 ~% ~{~<~% ~1:;~S~>~^ ~}"
232 "~:[This ~(~A~) is~;These ~(~A~)s are~] undefined:~
233 ~% ~{~<~% ~1:;~S~>~^ ~}"
245 ~[~:;~:*~& caught ~W fatal ERROR condition~:P~]~
246 ~[~:;~:*~& caught ~W ERROR condition~:P~]~
247 ~[~:;~:*~& caught ~W WARNING condition~:P~]~
248 ~[~:;~:*~& caught ~W STYLE-WARNING condition~:P~]~
249 ~[~:;~:*~& printed ~W note~:P~]"
250 abort-p
258 ;;; Evaluate BODY, then return (VALUES BODY-VALUE WARNINGS-P
259 ;;; FAILURE-P), where BODY-VALUE is the first value of the body, and
260 ;;; WARNINGS-P and FAILURE-P are as in CL:COMPILE or CL:COMPILE-FILE.
261 ;;; This also wraps up WITH-IR1-NAMESPACE functionality.
269 \f
270 ;;;; component compilation
273 #!+sb-doc
274 "The upper limit on the number of times that we will consecutively do IR1
275 optimization that doesn't introduce any new code. A finite limit is
281 ;;; Repeatedly optimize COMPONENT until no further optimizations can
282 ;;; be found or we hit our iteration limit. When we hit the limit, we
283 ;;; clear the component and block REOPTIMIZE flags to discourage the
284 ;;; next optimization attempt from pounding on the same code.
324 ;;; KLUDGE: This was bumped from 5 to 10 in a DTC patch ported by MNA
325 ;;; from CMU CL into sbcl-0.6.11.44, the same one which allowed IR1
326 ;;; transforms to be delayed. Either DTC or MNA or both didn't explain
327 ;;; why, and I don't know what the rationale was. -- WHN 2001-04-28
328 ;;;
329 ;;; FIXME: It would be good to document why it's important to have a
330 ;;; large value here, and what the drawbacks of an excessively large
331 ;;; value are; and it might also be good to make it depend on
332 ;;; optimization policy.
338 ;;; Iterate doing FIND-DFO until no new dead code is discovered.
351 ;;; Do all the IR1 phases for a non-top-level component.
377 ;; We delay the generation of type checks until
378 ;; the type constraints have had time to
379 ;; propagate, else the compiler can confuse itself.
409 ;; Assign BLOCK-NUMBER for any cleanup blocks introduced by
410 ;; stack analysis. There shouldn't be any unreachable code after
411 ;; control, so this won't delete anything.
459 #-sb-xc-host
471 code-length
472 trace-table
473 fixup-notes
479 code-length
480 trace-table
481 fixup-notes
485 ;; We're done, so don't bother keeping anything around.
490 ;;; Delete components with no external entry points before we try to
491 ;;; generate code. Unreachable closures can cause IR2 conversion to
492 ;;; puke on itself, since it is the reference to the closure which
493 ;;; normally causes the components to be combined.
500 (:external
508 ;; miscellaneous sanity checks
509 ;;
510 ;; FIXME: These are basically pretty wimpy compared to the checks done
511 ;; by the old CHECK-IR1-CONSISTENCY code. It would be really nice to
512 ;; make those internal consistency checks work again and use them.
517 ;; sanity check to prevent weirdness from propagating insidiously as
518 ;; far from its root cause as it did in bug 138: Make sure that
519 ;; thing-to-COMPONENT links are consistent.
534 #|
546 |#
548 ;; FIXME: What is MAYBE-MUMBLE for? Do we need it any more?
565 \f
566 ;;;; clearing global data structures
567 ;;;;
568 ;;;; FIXME: Is it possible to get rid of this stuff, getting rid of
569 ;;;; global data structures entirely when possible and consing up the
570 ;;;; others from scratch instead of clearing and reusing them?
572 ;;; Clear the INFO in constants in the *FREE-VARS*, etc. In
573 ;;; addition to allowing stuff to be reclaimed, this is required for
574 ;;; correct assignment of constant offsets, since we need to assign a
575 ;;; new offset for each component. We don't clear the FUNCTIONAL-INFO
576 ;;; slots, since they are used to keep track of functions across
577 ;;; component boundaries.
590 ;;; Blow away the REFS for all global variables, and let COMPONENT
591 ;;; be recycled.
603 x))
611 ;;; Clear global variables used by the compiler.
612 ;;;
613 ;;; FIXME: It seems kinda nasty and unmaintainable to have to do this,
614 ;;; and it adds overhead even when people aren't using the compiler.
615 ;;; Perhaps we could make these global vars unbound except when
616 ;;; actually in use, so that this function could go away.
619 ;; Clear global tables.
625 ;; Clear debug counters and tables.
641 ;; (Note: The CMU CL code used to set CL::*GENSYM-COUNTER* to zero here.
642 ;; Superficially, this seemed harmful -- the user could reasonably be
643 ;; surprised if *GENSYM-COUNTER* turned back to zero when something was
644 ;; compiled. A closer inspection showed that this actually turned out to be
645 ;; harmless in practice, because CLEAR-STUFF was only called from within
646 ;; forms which bound CL::*GENSYM-COUNTER* to zero. However, this means that
647 ;; even though zeroing CL::*GENSYM-COUNTER* here turned out to be harmless in
648 ;; practice, it was also useless in practice. So we don't do it any more.)
651 \f
652 ;;;; trace output
654 ;;; Print out some useful info about COMPONENT to STREAM.
675 \f
676 ;;;; file reading
677 ;;;;
678 ;;;; When reading from a file, we have to keep track of some source
679 ;;;; information. We also exploit our ability to back up for printing
680 ;;;; the error context and for recovering from errors.
681 ;;;;
682 ;;;; The interface we provide to this stuff is the stream-oid
683 ;;;; SOURCE-INFO structure. The bookkeeping is done as a side effect
684 ;;;; of getting the next source form.
686 ;;; A FILE-INFO structure holds all the source information for a
687 ;;; given file.
689 ;; If a file, the truename of the corresponding source file. If from
690 ;; a Lisp form, :LISP. If from a stream, :STREAM.
692 ;; the defaulted, but not necessarily absolute file name (i.e. prior
693 ;; to TRUENAME call.) Null if not a file. This is used to set
694 ;; *COMPILE-FILE-PATHNAME*, and if absolute, is dumped in the
695 ;; debug-info.
697 ;; the file's write date (if relevant)
699 ;; the source path root number of the first form in this file (i.e.
700 ;; the total number of forms converted previously in this
701 ;; compilation)
703 ;; parallel vectors containing the forms read out of the file and
704 ;; the file positions that reading of each form started at (i.e. the
705 ;; end of the previous form)
709 ;;; The SOURCE-INFO structure provides a handle on all the source
710 ;;; information for an entire compilation.
712 #-no-ansi-print-object
716 ;; the UT that compilation started at
718 ;; the FILE-INFO structure for this compilation
720 ;; the stream that we are using to read the FILE-INFO, or NIL if
721 ;; no stream has been opened yet
724 ;;; Given a pathname, return a SOURCE-INFO structure.
727 :untruename file
732 ;;; Return a SOURCE-INFO to describe the incremental compilation of FORM.
739 ;;; Return a SOURCE-INFO which will read from STREAM.
745 ;;; Return a form read from STREAM; or for EOF use the trick,
746 ;;; popularized by Kent Pitman, of returning STREAM itself. If an
747 ;;; error happens, then convert it to standard abort-the-compilation
748 ;;; error condition (possibly recording some extra location
749 ;;; information).
754 :condition condition
755 ;; We don't need to supply :POSITION here because
756 ;; READER-ERRORs already know their position in the file.
757 ))
758 ;; ANSI, in its wisdom, says that READ should return END-OF-FILE
759 ;; (and that this is not a READER-ERROR) when it encounters end of
760 ;; file in the middle of something it's trying to read.
763 :condition condition
764 ;; We need to supply :POSITION here because the END-OF-FILE
765 ;; condition doesn't carry the position that the user
766 ;; probably cares about, where the failed READ began.
769 ;;; If STREAM is present, return it, otherwise open a stream to the
770 ;;; current file. There must be a current file.
771 ;;;
772 ;;; FIXME: This is probably an unnecessarily roundabout way to do
773 ;;; things now that we process a single file in COMPILE-FILE (unlike
774 ;;; the old CMU CL code, which accepted multiple files). Also, the old
775 ;;; comment said
776 ;;; When we open a new file, we also reset *PACKAGE* and policy.
777 ;;; This gives the effect of rebinding around each file.
778 ;;; which doesn't seem to be true now. Check to make sure that if
779 ;;; such rebinding is necessary, it's still done somewhere.
789 ;;; Close the stream in INFO if it is open.
797 ;;; Read and compile the source file.
814 nil)))))))
816 ;;; Return the INDEX'th source form read from INFO and the position
817 ;;; where it was read.
823 \f
824 ;;;; processing of top level forms
826 ;;; This is called by top level form processing when we are ready to
827 ;;; actually compile something. If *BLOCK-COMPILE* is T, then we still
828 ;;; convert the form, but delay compilation, pushing the result on
829 ;;; *TOPLEVEL-LAMBDAS* instead.
839 ;;; Macroexpand FORM in the current environment with an error handler.
840 ;;; We only expand one level, so that we retain all the intervening
841 ;;; forms in the source path.
849 condition))))
851 ;;; Process a PROGN-like portion of a top level form. FORMS is a list of
852 ;;; the forms, and PATH is the source path of the FORM they came out of.
853 ;;; COMPILE-TIME-TOO is as in ANSI "3.2.3.1 Processing of Top Level Forms".
859 ;;; Process a top level use of LOCALLY, or anything else (e.g.
860 ;;; MACROLET) at top level which has declarations and ordinary forms.
861 ;;; We parse declarations and then recursively process the body.
867 ;; FIXME: VALUES declaration
868 ;;
869 ;; Binding *POLICY* is pretty much of a hack, since it
870 ;; causes LOCALLY to "capture" enclosed proclamations. It
871 ;; is necessary because CONVERT-AND-MAYBE-COMPILE uses the
872 ;; value of *POLICY* as the policy. The need for this hack
873 ;; is due to the quirk that there is no way to represent in
874 ;; a POLICY that an optimize quality came from the default.
875 ;;
876 ;; FIXME: Ideally, something should be done so that DECLAIM
877 ;; inside LOCALLY works OK. Failing that, at least we could
878 ;; issue a warning instead of silently screwing up.
880 ;; This is probably also a hack
882 ;; ditto
886 ;;; Parse an EVAL-WHEN situations list, returning three flags,
887 ;;; (VALUES COMPILE-TOPLEVEL LOAD-TOPLEVEL EXECUTE), indicating
888 ;;; the types of situations present in the list.
892 '(:compile-toplevel
893 compile
894 :load-toplevel
895 load
896 :execute
897 eval)))
902 deprecated-names)))
904 situations)
909 ;;; utilities for extracting COMPONENTs of FUNCTIONALs
920 result))))
924 result))))
927 &key
928 name
930 ;; I'd thought NIL should
931 ;; work, but it doesn't.
932 ;; -- WHN 2001-09-20
942 definition
944 name))))
954 fun)))
956 ;;; Compile LAMBDA-EXPRESSION into *COMPILE-OBJECT*, returning a
957 ;;; description of the result.
958 ;;; * If *COMPILE-OBJECT* is a CORE-OBJECT, then write the function
959 ;;; into core and return the compiled FUNCTION value.
960 ;;; * If *COMPILE-OBJECT* is a fasl file, then write the function
961 ;;; into the fasl file and return a dump handle.
962 ;;;
963 ;;; If NAME is provided, then we try to use it as the name of the
964 ;;; function for debugging/diagnostic information.
967 &key
968 name
970 ;; This magical idiom seems to be the appropriate
971 ;; path for compiling standalone LAMBDAs, judging
972 ;; from the CMU CL code and experiment, so it's a
973 ;; nice default for things where we don't have a
974 ;; real source path (as in e.g. inside CL:COMPILE).
983 :name name
986 ;; FIXME: The compile-it code from here on is sort of a
987 ;; twisted version of the code in COMPILE-TOPLEVEL. It'd be
988 ;; better to find a way to share the code there; or
989 ;; alternatively, to use this code to replace the code there.
990 ;; (The second alternative might be pretty easy if we used
991 ;; the :LOCALL-ONLY option to IR1-FOR-LAMBDA. Then maybe the
992 ;; whole FUNCTIONAL-KIND=:TOPLEVEL case could go away..)
1000 ;; FIXME: This is more monkey see monkey do based on CMU CL
1001 ;; code. If anyone figures out why to only prescan HAIRY-TOP
1002 ;; and TOP-COMPONENTS here, instead of *ALL-COMPONENTS* or
1003 ;; some other combination of results from FIND-INITIAL-VALUES,
1004 ;; it'd be good to explain it.
1020 result
1021 ;; KLUDGE: This code duplicates some other code in this
1022 ;; file. In the great reorganzation, the flow of program
1023 ;; logic changed from the original CMUCL model, and that
1024 ;; path (as of sbcl-0.7.5 in SUB-COMPILE-FILE) was no
1025 ;; longer followed for CORE-OBJECTS, leading to BUG
1026 ;; 156. This place is transparently not the right one for
1027 ;; this code, but I don't have a clear enough overview of
1028 ;; the compiler to know how to rearrange it all so that
1029 ;; this operation fits in nicely, and it was blocking
1030 ;; reimplementation of (DECLAIM (INLINE FOO)) (MACROLET
1031 ;; ((..)) (DEFUN FOO ...))
1032 ;;
1033 ;; FIXME: This KLUDGE doesn't solve all the problem in an
1034 ;; ideal way, as (1) definitions typed in at the REPL
1035 ;; without an INLINE declaration will give a NULL
1036 ;; FUNCTION-LAMBDA-EXPRESSION (allowable, but not ideal)
1037 ;; and (2) INLINE declarations will yield a
1038 ;; FUNCTION-LAMBDA-EXPRESSION headed by
1039 ;; SB-C:LAMBDA-WITH-LEXENV, even for null LEXENV. -- CSR,
1040 ;; 2002-07-02
1041 ;;
1042 ;; (2) is probably fairly easy to fix -- it is, after all,
1043 ;; a matter of list manipulation (or possibly of teaching
1044 ;; CL:FUNCTION about SB-C:LAMBDA-WITH-LEXENV). (1) is
1045 ;; significantly harder, as the association between
1046 ;; function object and source is a tricky one.
1047 ;;
1048 ;; FUNCTION-LAMBDA-EXPRESSION "works" (i.e. returns a
1049 ;; non-NULL list) when the function in question has been
1050 ;; compiled by (COMPILE <x> '(LAMBDA ...)); it does not
1051 ;; work when it has been compiled as part of the top-level
1052 ;; EVAL strategy of compiling everything inside (LAMBDA ()
1053 ;; ...). -- CSR, 2002-11-02
1065 (%compile lambda-expression
1067 :name name
1072 ;;; Process a top level FORM with the specified source PATH.
1073 ;;; * If this is a magic top level form, then do stuff.
1074 ;;; * If this is a macro, then expand it.
1075 ;;; * Otherwise, just compile it.
1076 ;;;
1077 ;;; COMPILE-TIME-TOO is as defined in ANSI
1078 ;;; "3.2.3.1 Processing of Top Level Forms".
1089 path)
1093 ;; When we're cross-compiling, consider: what should we
1094 ;; do when we hit e.g.
1095 ;; (EVAL-WHEN (:COMPILE-TOPLEVEL)
1096 ;; (DEFUN FOO (X) (+ 7 X)))?
1097 ;; DEFUN has a macro definition in the cross-compiler,
1098 ;; and a different macro definition in the target
1099 ;; compiler. The only sensible thing is to use the
1100 ;; target compiler's macro definition, since the
1101 ;; cross-compiler's macro is in general into target
1102 ;; functions which can't meaningfully be executed at
1103 ;; cross-compilation time. So make sure we do the EVAL
1104 ;; here, before we macroexpand.
1105 ;;
1106 ;; Then things get even dicier with something like
1107 ;; (DEFCONSTANT-EQX SB!XC:LAMBDA-LIST-KEYWORDS ..)
1108 ;; where we have to make sure that we don't uncross
1109 ;; the SB!XC: prefix before we do EVAL, because otherwise
1110 ;; we'd be trying to redefine the cross-compilation host's
1111 ;; constants.
1112 ;;
1113 ;; (Isn't it fun to cross-compile Common Lisp?:-)
1114 #+sb-xc-host
1119 ;; like SB!XC:DEFCONSTANT and SB!XC:DEFTYPE
1120 ;; should be equivalent to their CL: counterparts
1121 ;; when being compiled as target code. We leave
1122 ;; the rest of the form uncrossed because macros
1123 ;; might yet expand into EVAL-WHEN stuff, and
1124 ;; things inside EVAL-WHEN can't be uncrossed
1125 ;; until after we've EVALed them in the
1126 ;; cross-compilation host.)
1130 slightly-uncrossed)))
1132 ;; (Now that we're no longer processing toplevel
1133 ;; forms, and hence no longer need to worry about
1134 ;; EVAL-WHEN, we can uncross everything.)
1136 ;; (We have to demote COMPILE-TIME-TOO to NIL
1137 ;; here, no matter what it was before, since
1138 ;; otherwise we'd tend to EVAL subforms more than
1139 ;; once, because of WHEN COMPILE-TIME-TOO form
1140 ;; above.)
1142 ;; When we're not cross-compiling, we only need to
1143 ;; macroexpand once, so we can follow the 1-thru-6
1144 ;; sequence of steps in ANSI's "3.2.3.1 Processing of
1145 ;; Top Level Forms".
1146 #-sb-xc-host
1154 path
1155 compile-time-too))))))
1157 #+sb-xc-host
1158 ;; (There are no xc EVAL-WHEN issues in the ATOM case until
1159 ;; (1) SBCL gets smart enough to handle global
1160 ;; DEFINE-SYMBOL-MACRO or SYMBOL-MACROLET and (2) SBCL
1161 ;; implementors start using symbol macros in a way which
1162 ;; interacts with SB-XC/CL distinction.)
1164 #-sb-xc-host
1170 form))))
1172 ;; In the cross-compiler, top level COLD-FSET arranges
1173 ;; for static linking at cold init time.
1174 #+sb-xc-host
1180 lambda-expression
1181 path)))
1187 ;; CT, LT, and E here are as in Figure 3-7 of ANSI
1188 ;; "3.2.3.1 Processing of Top Level Forms".
1193 e))))
1195 body path new-compile-time-too))
1201 magic
1206 path
1207 compile-time-too))
1211 magic
1215 path
1216 compile-time-too
1226 \f
1227 ;;;; load time value support
1228 ;;;;
1229 ;;;; (See EMIT-MAKE-LOAD-FORM.)
1231 ;;; Return T if we are currently producing a fasl file and hence
1232 ;;; constants need to be dumped carefully.
1236 ;;; Compile FORM and arrange for it to be called at load-time. Return
1237 ;;; the dumper handle and our best guess at the type of the object.
1247 ;;; Compile the FORMS and arrange for them to be called (for effect,
1248 ;;; not value) at load time.
1253 ;;; Do the actual work of COMPILE-LOAD-TIME-VALUE or
1254 ;;; COMPILE-MAKE-LOAD-FORM-INIT-FORMS.
1260 lambda)))
1262 ;;; This is called by COMPILE-TOPLEVEL when it was passed T for
1263 ;;; LOAD-TIME-VALUE-P (which happens in COMPILE-LOAD-TIME-STUFF). We
1264 ;;; don't try to combine this component with anything else and frob
1265 ;;; the name. If not in a :TOPLEVEL component, then don't bother
1266 ;;; compiling, because it was merged with a run-time component.
1275 \f
1276 ;;;; COMPILE-FILE
1286 ;;; Smash LAMBDAS into a single component, compile it, and arrange for
1287 ;;; the resulting function to be called.
1297 ;;; Compile top level code and call the top level lambdas. We pick off
1298 ;;; top level lambdas in non-top-level components here, calling
1299 ;;; SUB-c-t-l-l on each subsequence of normal top level lambdas.
1308 lambdas
1309 ;; this used to read ":start start", but
1310 ;; start can be greater than len, which
1311 ;; is an error according to ANSI - CSR,
1312 ;; 2002-04-25
1314 len)))
1323 ;;; Compile LAMBDAS (a list of CLAMBDAs for top level forms) into the
1324 ;;; object file.
1325 ;;;
1326 ;;; LOAD-TIME-VALUE-P seems to control whether it's MAKE-LOAD-FORM and
1327 ;;; COMPILE-LOAD-TIME-VALUE stuff. -- WHN 20000201
1361 ;;; Actually compile any stuff that has been queued up for block
1362 ;;; compilation.
1380 lexenv))
1384 nil
1400 lexenv))
1410 ;;; Read all forms from INFO and compile them, with output to OBJECT.
1411 ;;; Return (VALUES NIL WARNINGS-P FAILURE-P).
1439 ;; FIXME: Do we need this rebinding here? It's a literal
1440 ;; translation of the old CMU CL rebinding to
1441 ;; (OR *BACKEND-INFO-ENVIRONMENT* *INFO-ENVIRONMENT*),
1442 ;; and it's not obvious whether the rebinding to itself is
1443 ;; needed that SBCL doesn't need *BACKEND-INFO-ENVIRONMENT*.
1460 nil)))
1461 ;; Some errors are sufficiently bewildering that we just fail
1462 ;; immediately, without trying to recover and compile more of
1463 ;; the input file.
1467 "~@<compilation aborted because of fatal error: ~2I~_~A~:>"
1468 condition)
1471 ;;; Return a pathname for the named file. The file must exist.
1480 new
1493 ;;; Print some junk at the beginning and end of compilation.
1501 file-info)
1503 :print-weekday nil
1509 won
1515 ;;; Open some files and call SUB-COMPILE-FILE. If something unwinds
1516 ;;; out of the compile, then abort the writing of the output file, so
1517 ;;; that we don't overwrite it with known garbage.
1520 &key
1522 ;; ANSI options
1524 ;; FIXME: ANSI doesn't seem to say anything about
1525 ;; *COMPILE-VERBOSE* and *COMPILE-PRINT* being rebound by this
1526 ;; function..
1531 ;; extensions
1535 #!+sb-doc
1536 "Compile INPUT-FILE, producing a corresponding fasl file and returning
1537 its filename. Besides the ANSI &KEY arguments :OUTPUT-FILE, :VERBOSE,
1538 :PRINT, and :EXTERNAL-FORMAT,the following extensions are supported:
1539 :TRACE-FILE
1540 If given, internal data structures are dumped to the specified
1541 file, or if a value of T is given, to a file of *.trace type
1542 derived from the input file name.
1543 Also, as a workaround for vaguely-non-ANSI behavior, the :BLOCK-COMPILE
1544 argument is quasi-supported, to determine whether multiple
1545 functions are compiled together as a unit, resolving function
1546 references at compile time. NIL means that global function names
1547 are never resolved at compilation time. Currently NIL is the
1550 circumstances, the compiler's runtime scales badly when it
1551 tries to do this for large files. If/when this performance
1552 problem is fixed, the block compilation default behavior will
1553 probably be made dependent on the SPEED and COMPILATION-SPEED
1554 optimization values, and the :BLOCK-COMPILE argument will probably
1555 become deprecated."
1582 default-trace-file-pathname
1584 default-trace-file-pathname))))
1593 dummy)
1614 ;; Hack around filesystem race condition...
1616 nil)
1617 warnings-p
1618 failure-p)))
1619 \f
1620 ;;; a helper function for COMPILE-FILE-PATHNAME: the default for
1621 ;;; the OUTPUT-FILE argument
1622 ;;;
1623 ;;; ANSI: The defaults for the OUTPUT-FILE are taken from the pathname
1624 ;;; that results from merging the INPUT-FILE with the value of
1625 ;;; *DEFAULT-PATHNAME-DEFAULTS*, except that the type component should
1626 ;;; default to the appropriate implementation-defined default type for
1627 ;;; compiled files.
1631 retyped))
1633 ;;; KLUDGE: Part of the ANSI spec for this seems contradictory:
1634 ;;; If INPUT-FILE is a logical pathname and OUTPUT-FILE is unsupplied,
1635 ;;; the result is a logical pathname. If INPUT-FILE is a logical
1636 ;;; pathname, it is translated into a physical pathname as if by
1637 ;;; calling TRANSLATE-LOGICAL-PATHNAME.
1638 ;;; So I haven't really tried to make this precisely ANSI-compatible
1639 ;;; at the level of e.g. whether it returns logical pathname or a
1640 ;;; physical pathname. Patches to make it more correct are welcome.
1641 ;;; -- WHN 2000-12-09
1643 &key
1645 input-file))
1647 #!+sb-doc
1648 "Return a pathname describing what file COMPILE-FILE would write to given
1649 these arguments."
1651 \f
1652 ;;;; MAKE-LOAD-FORM stuff
1654 ;;; The entry point for MAKE-LOAD-FORM support. When IR1 conversion
1655 ;;; finds a constant structure, it invokes this to arrange for proper
1656 ;;; dumping. If it turns out that the constant has already been
1657 ;;; dumped, then we don't need to do anything.
1658 ;;;
1659 ;;; If the constant hasn't been dumped, then we check to see whether
1660 ;;; we are in the process of creating it. We detect this by
1661 ;;; maintaining the special *CONSTANTS-BEING-CREATED* as a list of all
1662 ;;; the constants we are in the process of creating. Actually, each
1663 ;;; entry is a list of the constant and any init forms that need to be
1664 ;;; processed on behalf of that constant.
1665 ;;;
1666 ;;; It's not necessarily an error for this to happen. If we are
1667 ;;; processing the init form for some object that showed up *after*
1668 ;;; the original reference to this constant, then we just need to
1669 ;;; defer the processing of that init form. To detect this, we
1670 ;;; maintain *CONSTANTS-CREATED-SINCE-LAST-INIT* as a list of the
1671 ;;; constants created since the last time we started processing an
1672 ;;; init form. If the constant passed to emit-make-load-form shows up
1673 ;;; in this list, then there is a circular chain through creation
1674 ;;; forms, which is an error.
1675 ;;;
1676 ;;; If there is some intervening init form, then we blow out of
1677 ;;; processing it by throwing to the tag PENDING-INIT. The value we
1678 ;;; throw is the entry from *CONSTANTS-BEING-CREATED*. This is so the
1679 ;;; offending init form can be tacked onto the init forms for the
1680 ;;; circular object.
1681 ;;;
1682 ;;; If the constant doesn't show up in *CONSTANTS-BEING-CREATED*, then
1683 ;;; we have to create it. We call MAKE-LOAD-FORM and check to see
1684 ;;; whether the creation form is the magic value
1685 ;;; :SB-JUST-DUMP-IT-NORMALLY. If it is, then we don't do anything. The
1686 ;;; dumper will eventually get its hands on the object and use the
1687 ;;; normal structure dumping noise on it.
1688 ;;;
1689 ;;; Otherwise, we bind *CONSTANTS-BEING-CREATED* and
1690 ;;; *CONSTANTS-CREATED-SINCE- LAST-INIT* and compile the creation form
1691 ;;; much the way LOAD-TIME-VALUE does. When this finishes, we tell the
1692 ;;; dumper to use that result instead whenever it sees this constant.
1693 ;;;
1694 ;;; Now we try to compile the init form. We bind
1695 ;;; *CONSTANTS-CREATED-SINCE-LAST-INIT* to NIL and compile the init
1696 ;;; form (and any init forms that were added because of circularity
1697 ;;; detection). If this works, great. If not, we add the init forms to
1698 ;;; the init forms for the object that caused the problems and let it
1699 ;;; deal with it.
1702 ;;; FIXME: Shouldn't these^ variables be unbound outside LET forms?
1706 ;; KLUDGE: This special hack is because I was too lazy
1707 ;; to rework DEF!STRUCT so that the MAKE-LOAD-FORM
1708 ;; function of LAYOUT returns nontrivial forms when
1709 ;; building the cross-compiler but :IGNORE-IT when
1710 ;; cross-compiling or running under the target Lisp. --
1711 ;; WHN 19990914
1724 (:sb-just-dump-it-normally
1726 t)
1727 (:ignore-it
1728 nil)
1745 constant
1747 creation-form)
1749 nil)
1751 constant)))
1757 collect name into names
1758 collect form into forms
1760 nil)))
1765 \f
1766 ;;;; Host compile time definitions
1767 #+sb-xc-host
1772 #+sb-xc-host