| blob | c7f19ce0ecdcc41fa5f31e1ee571dc721e55f7fe |
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?
35 ;;; Whether reference to a thing which cannot be defined causes a full
36 ;;; 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.)
64 ;;; The current non-macroexpanded toplevel form as printed when
65 ;;; *compile-print* is true.
69 #!+sb-doc
72 #!+sb-doc
75 #!+sb-doc
76 "When this is true, the compiler prints to *STANDARD-OUTPUT* progress
77 information about the phases of compilation of each function. (This
81 #!+sb-doc
82 "The defaulted pathname of the file currently being compiled, or NIL if not
85 #!+sb-doc
86 "The TRUENAME of the file currently being compiled, or NIL if not
93 ;;; the SOURCE-INFO structure for the current compilation. This is
94 ;;; null globally to indicate that we aren't currently in any
95 ;;; identifiable compilation.
98 ;;; This is true if we are within a WITH-COMPILATION-UNIT form (which
99 ;;; normally causes nested uses to be no-ops).
102 ;;; Count of the number of compilation units dynamically enclosed by
103 ;;; the current active WITH-COMPILATION-UNIT that were unwound out of.
106 ;;; Mumble conditional on *COMPILE-PROGRESS*.
123 ;; Used during compilation to map code paths to the matching
124 ;; instrumentation conses.
126 ;; Used during compilation to keep track of with source paths have been
127 ;; instrumented in which blocks.
129 ;; Stores the code coverage instrumentation results. Keys are namestrings,
130 ;; the value is a list of (CONS PATH STATE), where STATE is NIL for
131 ;; a path that has not been visited, and T for one that has.
134 \f
135 ;;;; WITH-COMPILATION-UNIT and WITH-COMPILATION-VALUES
138 #!+sb-doc
139 "Affects compilations that take place within its dynamic extent. It is
140 intended to be eg. wrapped around the compilation of all files in the same system.
142 Following options are defined:
144 :OVERRIDE Boolean-Form
145 One of the effects of this form is to delay undefined warnings until the
146 end of the form, instead of giving them at the end of each compilation.
147 If OVERRIDE is NIL (the default), then the outermost
148 WITH-COMPILATION-UNIT form grabs the undefined warnings. Specifying
149 OVERRIDE true causes that form to grab any enclosed warnings, even if it
150 is enclosed by another WITH-COMPILATION-UNIT.
152 :POLICY Optimize-Declaration-Form
153 Provides dynamic scoping for global compiler optimization qualities and
154 restrictions, limiting effects of subsequent OPTIMIZE proclamations and
155 calls to SB-EXT:RESTRICT-COMPILER-POLICY to the dynamic scope of BODY.
157 If OVERRIDE is false, specified POLICY is merged with current global
158 policy. If OVERRIDE is true, current global policy, including any
159 restrictions, is discarded in favor of the specified POLICY.
161 Supplying POLICY NIL is equivalent to the option not being supplied at
162 all, ie. dynamic scoping of policy does not take place.
164 This option is an SBCL-specific experimental extension: Interface
165 subject to change.
167 :SOURCE-NAMESTRING Namestring-Form
168 Attaches the value returned by the Namestring-Form to the internal
169 debug-source information as the namestring of the source file. Normally
170 the namestring of the input-file for COMPILE-FILE is used: this option
171 can be used to provide source-file information for functions compiled
172 using COMPILE, or to override the input-file of COMPILE-FILE.
174 If both an outer and an inner WITH-COMPILATION-UNIT provide a
175 SOURCE-NAMESTRING, the inner one takes precedence. Unaffected
176 by :OVERRIDE.
178 This is an SBCL-specific extension.
180 :SOURCE-PLIST Plist-Form
181 Attaches the value returned by the Plist-Form to internal debug-source
182 information of functions compiled in within the dynamic extent of BODY.
184 Primarily for use by development environments, in order to eg. associate
185 function definitions with editor-buffers. Can be accessed using
186 SB-INTROSPECT:DEFINITION-SOURCE-PLIST.
188 If an outer WITH-COMPILATION-UNIT form also provide a SOURCE-PLIST, it
189 is appended to the end of the provided SOURCE-PLIST. Unaffected
190 by :OVERRIDE.
192 This is an SBCL-specific extension.
194 Examples:
196 ;; Prevent proclamations from the file leaking, and restrict
197 ;; SAFETY to 3 -- otherwise uses the current global policy.
198 (with-compilation-unit (:policy '(optimize))
199 (restrict-compiler-policy 'safety 3)
202 ;; Using default policy instead of the current global one,
203 ;; except for DEBUG 3.
204 (with-compilation-unit (:policy '(optimize debug)
205 :override t)
208 ;; Same as if :POLICY had not been specified at all: SAFETY 3
209 ;; proclamation leaks out from WITH-COMPILATION-UNIT.
210 (with-compilation-unit (:policy nil)
211 (declaim (optimize safety))
213 "
226 ;; Inside another WITH-COMPILATION-UNIT, a WITH-COMPILATION-UNIT is
227 ;; ordinarily (unless OVERRIDE) basically a no-op.
256 ;;; Is NAME something that no conforming program can rely on
257 ;;; defining?
260 (:function
263 (:type
270 ;;; This is to be called at the end of a compilation unit. It signals
271 ;;; any residual warnings about unknown stuff, then prints the total
272 ;;; error counts. ABORT-P should be true when the compilation unit was
273 ;;; aborted by throwing out. ABORT-COUNT is the number of dynamically
274 ;;; enclosed nested compilation units that were aborted.
301 #+sb-xc-host nil
303 (:function
307 "~@<There is no function named ~S. References to ~S ~
308 in some contexts (like starts of blocks) have ~
309 special meaning, but here it would have to be a ~
310 function, and that shouldn't be right.~:@>" name
311 name))
314 "~@<The function ~S is undefined, and its name is ~
315 reserved by ANSI CL so that even if it were ~
316 defined later, the code doing so would not be ~
318 (:type
321 "~@<Undefined type ~S. The name starts with ~S: ~
322 probably use of a quoted type name in a context ~
323 where the name is not evaluated.~:@>"
326 "~@<Undefined type ~S. Note that name ~S is ~
327 reserved by ANSI CL, so code defining a type with ~
328 that name would not be portable.~:@>" name
329 name))))
338 "~W more use~:P of undefined ~(~A~) ~S"
339 more kind name)
341 "~W more use~:P of undefined ~(~A~) ~S"
342 more kind name))))))))))
352 abort-p)
356 "~& Undefined ~(~A~)~p:~
357 ~% ~{~<~% ~1:;~S~>~^ ~}"
360 ~[~:;~:*~& caught ~W ERROR condition~:P~]~
361 ~[~:;~:*~& caught ~W WARNING condition~:P~]~
362 ~[~:;~:*~& caught ~W STYLE-WARNING condition~:P~]~
363 ~[~:;~:*~& printed ~W note~:P~]"
372 ;;; Evaluate BODY, then return (VALUES BODY-VALUE WARNINGS-P
373 ;;; FAILURE-P), where BODY-VALUE is the first value of the body, and
374 ;;; WARNINGS-P and FAILURE-P are as in CL:COMPILE or CL:COMPILE-FILE.
375 ;;; This also wraps up WITH-IR1-NAMESPACE functionality.
383 \f
384 ;;;; component compilation
387 #!+sb-doc
388 "The upper limit on the number of times that we will consecutively do IR1
389 optimization that doesn't introduce any new code. A finite limit is
395 ;;; Repeatedly optimize COMPONENT until no further optimizations can
396 ;;; be found or we hit our iteration limit. When we hit the limit, we
397 ;;; clear the component and block REOPTIMIZE flags to discourage the
398 ;;; next optimization attempt from pounding on the same code.
442 ;;; KLUDGE: This was bumped from 5 to 10 in a DTC patch ported by MNA
443 ;;; from CMU CL into sbcl-0.6.11.44, the same one which allowed IR1
444 ;;; transforms to be delayed. Either DTC or MNA or both didn't explain
445 ;;; why, and I don't know what the rationale was. -- WHN 2001-04-28
446 ;;;
447 ;;; FIXME: It would be good to document why it's important to have a
448 ;;; large value here, and what the drawbacks of an excessively large
449 ;;; value are; and it might also be good to make it depend on
450 ;;; optimization policy.
456 ;;; Iterate doing FIND-DFO until no new dead code is discovered.
469 ;;; Do all the IR1 phases for a non-top-level component.
474 ;make this 0.
497 ;; We delay the generation of type checks until
498 ;; the type constraints have had time to
499 ;; propagate, else the compiler can confuse itself.
517 #!+inline-constants
519 #!+inline-constants
521 #!+inline-constants
535 ;; Assign BLOCK-NUMBER for any cleanup blocks introduced by
536 ;; stack analysis. There shouldn't be any unreachable code after
537 ;; control, so this won't delete anything.
587 #-sb-xc-host
599 code-length
600 trace-table
601 fixup-notes
607 code-length
608 trace-table
609 fixup-notes
613 ;; We're done, so don't bother keeping anything around.
618 ;;; Delete components with no external entry points before we try to
619 ;;; generate code. Unreachable closures can cause IR2 conversion to
620 ;;; puke on itself, since it is the reference to the closure which
621 ;;; normally causes the components to be combined.
628 (:external
636 ;; miscellaneous sanity checks
637 ;;
638 ;; FIXME: These are basically pretty wimpy compared to the checks done
639 ;; by the old CHECK-IR1-CONSISTENCY code. It would be really nice to
640 ;; make those internal consistency checks work again and use them.
645 ;; sanity check to prevent weirdness from propagating insidiously as
646 ;; far from its root cause as it did in bug 138: Make sure that
647 ;; thing-to-COMPONENT links are consistent.
653 ;; Record xref information before optimization. This way the
654 ;; stored xref data reflects the real source as closely as
655 ;; possible.
665 #|
677 |#
679 ;; FIXME: What is MAYBE-MUMBLE for? Do we need it any more?
693 \f
694 ;;;; clearing global data structures
695 ;;;;
696 ;;;; FIXME: Is it possible to get rid of this stuff, getting rid of
697 ;;;; global data structures entirely when possible and consing up the
698 ;;;; others from scratch instead of clearing and reusing them?
700 ;;; Clear the INFO in constants in the *FREE-VARS*, etc. In
701 ;;; addition to allowing stuff to be reclaimed, this is required for
702 ;;; correct assignment of constant offsets, since we need to assign a
703 ;;; new offset for each component. We don't clear the FUNCTIONAL-INFO
704 ;;; slots, since they are used to keep track of functions across
705 ;;; component boundaries.
720 ;;; Blow away the REFS for all global variables, and let COMPONENT
721 ;;; be recycled.
733 x))
741 ;;; Clear global variables used by the compiler.
742 ;;;
743 ;;; FIXME: It seems kinda nasty and unmaintainable to have to do this,
744 ;;; and it adds overhead even when people aren't using the compiler.
745 ;;; Perhaps we could make these global vars unbound except when
746 ;;; actually in use, so that this function could go away.
749 ;; Clear global tables.
755 ;; Clear debug counters and tables.
771 ;; (Note: The CMU CL code used to set CL::*GENSYM-COUNTER* to zero here.
772 ;; Superficially, this seemed harmful -- the user could reasonably be
773 ;; surprised if *GENSYM-COUNTER* turned back to zero when something was
774 ;; compiled. A closer inspection showed that this actually turned out to be
775 ;; harmless in practice, because CLEAR-STUFF was only called from within
776 ;; forms which bound CL::*GENSYM-COUNTER* to zero. However, this means that
777 ;; even though zeroing CL::*GENSYM-COUNTER* here turned out to be harmless in
778 ;; practice, it was also useless in practice. So we don't do it any more.)
781 \f
782 ;;;; trace output
784 ;;; Print out some useful info about COMPONENT to STREAM.
805 \f
806 ;;;; file reading
807 ;;;;
808 ;;;; When reading from a file, we have to keep track of some source
809 ;;;; information. We also exploit our ability to back up for printing
810 ;;;; the error context and for recovering from errors.
811 ;;;;
812 ;;;; The interface we provide to this stuff is the stream-oid
813 ;;;; SOURCE-INFO structure. The bookkeeping is done as a side effect
814 ;;;; of getting the next source form.
816 ;;; A FILE-INFO structure holds all the source information for a
817 ;;; given file.
820 #-no-ansi-print-object
824 ;; If a file, the truename of the corresponding source file. If from
825 ;; a Lisp form, :LISP. If from a stream, :STREAM.
827 ;; the external format that we'll call OPEN with, if NAME is a file.
829 ;; the defaulted, but not necessarily absolute file name (i.e. prior
830 ;; to TRUENAME call.) Null if not a file. This is used to set
831 ;; *COMPILE-FILE-PATHNAME*, and if absolute, is dumped in the
832 ;; debug-info.
834 ;; the file's write date (if relevant)
836 ;; the source path root number of the first form in this file (i.e.
837 ;; the total number of forms converted previously in this
838 ;; compilation)
840 ;; parallel vectors containing the forms read out of the file and
841 ;; the file positions that reading of each form started at (i.e. the
842 ;; end of the previous form)
846 ;;; The SOURCE-INFO structure provides a handle on all the source
847 ;;; information for an entire compilation.
849 #-no-ansi-print-object
853 ;; the UT that compilation started at
855 ;; the IRT that compilation started at
857 ;; the FILE-INFO structure for this compilation
859 ;; the stream that we are using to read the FILE-INFO, or NIL if
860 ;; no stream has been opened yet
862 ;; if the current compilation is recursive (e.g., due to EVAL-WHEN
863 ;; processing at compile-time), the invoking compilation's
864 ;; source-info.
867 ;;; Given a pathname, return a SOURCE-INFO structure.
872 :external-format external-format
875 ;;; Return a SOURCE-INFO to describe the incremental compilation of FORM.
883 ;;; Walk up the SOURCE-INFO list until we either reach a SOURCE-INFO
884 ;;; with no parent (e.g., from a REPL evaluation) or until we reach a
885 ;;; SOURCE-INFO whose FILE-INFO denotes a file.
893 finfo))))
895 ;;; Return a form read from STREAM; or for EOF use the trick,
896 ;;; popularized by Kent Pitman, of returning STREAM itself. If an
897 ;;; error happens, then convert it to standard abort-the-compilation
898 ;;; error condition (possibly recording some extra location
899 ;;; information).
905 ;; We don't need to supply :POSITION here because
906 ;; READER-ERRORs already know their position in the file.
907 :condition condition
909 ;; ANSI, in its wisdom, says that READ should return END-OF-FILE
910 ;; (and that this is not a READER-ERROR) when it encounters end of
911 ;; file in the middle of something it's trying to read.
914 :condition condition
915 ;; We need to supply :POSITION here because the END-OF-FILE
916 ;; condition doesn't carry the position that the user
917 ;; probably cares about, where the failed READ began.
918 :position position
922 :condition condition
923 :position position
926 ;;; If STREAM is present, return it, otherwise open a stream to the
927 ;;; current file. There must be a current file.
928 ;;;
929 ;;; FIXME: This is probably an unnecessarily roundabout way to do
930 ;;; things now that we process a single file in COMPILE-FILE (unlike
931 ;;; the old CMU CL code, which accepted multiple files). Also, the old
932 ;;; comment said
933 ;;; When we open a new file, we also reset *PACKAGE* and policy.
934 ;;; This gives the effect of rebinding around each file.
935 ;;; which doesn't seem to be true now. Check to make sure that if
936 ;;; such rebinding is necessary, it's still done somewhere.
949 ;;; Close the stream in INFO if it is open.
957 ;;; Loop over FORMS retrieved from INFO. Used by COMPILE-FILE and
958 ;;; LOAD when loading from a FILE-STREAM associated with a source
959 ;;; file.
975 file-info))))
978 file-info))
982 ;;; Read and compile the source file.
989 ;;; Return the INDEX'th source form read from INFO and the position
990 ;;; where it was read.
996 \f
997 ;;;; processing of top level forms
999 ;;; This is called by top level form processing when we are ready to
1000 ;;; actually compile something. If *BLOCK-COMPILE* is T, then we still
1001 ;;; convert the form, but delay compilation, pushing the result on
1002 ;;; *TOPLEVEL-LAMBDAS* instead.
1006 ;; Don't bother to compile simple objects that just sit there.
1019 nil)))))
1021 ;;; Macroexpand FORM in the current environment with an error handler.
1022 ;;; We only expand one level, so that we retain all the intervening
1023 ;;; forms in the source path.
1031 condition))))
1033 ;;; Process a PROGN-like portion of a top level form. FORMS is a list of
1034 ;;; the forms, and PATH is the source path of the FORM they came out of.
1035 ;;; COMPILE-TIME-TOO is as in ANSI "3.2.3.1 Processing of Top Level Forms".
1041 ;;; Process a top level use of LOCALLY, or anything else (e.g.
1042 ;;; MACROLET) at top level which has declarations and ordinary forms.
1043 ;;; We parse declarations and then recursively process the body.
1049 ;; FIXME: VALUES declaration
1050 ;;
1051 ;; Binding *POLICY* is pretty much of a hack, since it
1052 ;; causes LOCALLY to "capture" enclosed proclamations. It
1053 ;; is necessary because CONVERT-AND-MAYBE-COMPILE uses the
1054 ;; value of *POLICY* as the policy. The need for this hack
1055 ;; is due to the quirk that there is no way to represent in
1056 ;; a POLICY that an optimize quality came from the default.
1057 ;;
1058 ;; FIXME: Ideally, something should be done so that DECLAIM
1059 ;; inside LOCALLY works OK. Failing that, at least we could
1060 ;; issue a warning instead of silently screwing up.
1062 ;; This is probably also a hack
1064 ;; ditto
1068 ;;; Parse an EVAL-WHEN situations list, returning three flags,
1069 ;;; (VALUES COMPILE-TOPLEVEL LOAD-TOPLEVEL EXECUTE), indicating
1070 ;;; the types of situations present in the list.
1074 '(:compile-toplevel
1075 compile
1076 :load-toplevel
1077 load
1078 :execute
1079 eval)))
1084 deprecated-names)))
1086 situations)
1091 ;;; utilities for extracting COMPONENTs of FUNCTIONALs
1102 result))))
1106 result))))
1109 &key
1110 name
1112 ;; I'd thought NIL should
1113 ;; work, but it doesn't.
1114 ;; -- WHN 2001-09-20
1125 lambda-expression
1127 ;; Convert the XEP using the policy of the real function. Otherwise
1128 ;; the wrong policy will be used for deciding whether to type-check
1129 ;; the parameters of the real function (via CONVERT-CALL /
1130 ;; PROPAGATE-TO-ARGS). -- JES, 2007-02-27
1133 :source-name source-name
1145 ;; Any leftover REFs to FUN outside local calls get replaced with the
1146 ;; XEP.
1153 xep
1154 fun)
1155 xep)))
1157 ;;; Compile LAMBDA-EXPRESSION into *COMPILE-OBJECT*, returning a
1158 ;;; description of the result.
1159 ;;; * If *COMPILE-OBJECT* is a CORE-OBJECT, then write the function
1160 ;;; into core and return the compiled FUNCTION value.
1161 ;;; * If *COMPILE-OBJECT* is a fasl file, then write the function
1162 ;;; into the fasl file and return a dump handle.
1163 ;;;
1164 ;;; If NAME is provided, then we try to use it as the name of the
1165 ;;; function for debugging/diagnostic information.
1168 &key
1169 name
1171 ;; This magical idiom seems to be the appropriate
1172 ;; path for compiling standalone LAMBDAs, judging
1173 ;; from the CMU CL code and experiment, so it's a
1174 ;; nice default for things where we don't have a
1175 ;; real source path (as in e.g. inside CL:COMPILE).
1185 :name name
1188 ;; FIXME: The compile-it code from here on is sort of a
1189 ;; twisted version of the code in COMPILE-TOPLEVEL. It'd be
1190 ;; better to find a way to share the code there; or
1191 ;; alternatively, to use this code to replace the code there.
1192 ;; (The second alternative might be pretty easy if we used
1193 ;; the :LOCALL-ONLY option to IR1-FOR-LAMBDA. Then maybe the
1194 ;; whole FUNCTIONAL-KIND=:TOPLEVEL case could go away..)
1212 result
1213 ;; KLUDGE: This code duplicates some other code in this
1214 ;; file. In the great reorganzation, the flow of program
1215 ;; logic changed from the original CMUCL model, and that
1216 ;; path (as of sbcl-0.7.5 in SUB-COMPILE-FILE) was no
1217 ;; longer followed for CORE-OBJECTS, leading to BUG
1218 ;; 156. This place is transparently not the right one for
1219 ;; this code, but I don't have a clear enough overview of
1220 ;; the compiler to know how to rearrange it all so that
1221 ;; this operation fits in nicely, and it was blocking
1222 ;; reimplementation of (DECLAIM (INLINE FOO)) (MACROLET
1223 ;; ((..)) (DEFUN FOO ...))
1224 ;;
1225 ;; FIXME: This KLUDGE doesn't solve all the problem in an
1226 ;; ideal way, as (1) definitions typed in at the REPL
1227 ;; without an INLINE declaration will give a NULL
1228 ;; FUNCTION-LAMBDA-EXPRESSION (allowable, but not ideal)
1229 ;; and (2) INLINE declarations will yield a
1230 ;; FUNCTION-LAMBDA-EXPRESSION headed by
1231 ;; SB-C:LAMBDA-WITH-LEXENV, even for null LEXENV. -- CSR,
1232 ;; 2002-07-02
1233 ;;
1234 ;; (2) is probably fairly easy to fix -- it is, after all,
1235 ;; a matter of list manipulation (or possibly of teaching
1236 ;; CL:FUNCTION about SB-C:LAMBDA-WITH-LEXENV). (1) is
1237 ;; significantly harder, as the association between
1238 ;; function object and source is a tricky one.
1239 ;;
1240 ;; FUNCTION-LAMBDA-EXPRESSION "works" (i.e. returns a
1241 ;; non-NULL list) when the function in question has been
1242 ;; compiled by (COMPILE <x> '(LAMBDA ...)); it does not
1243 ;; work when it has been compiled as part of the top-level
1244 ;; EVAL strategy of compiling everything inside (LAMBDA ()
1245 ;; ...). -- CSR, 2002-11-02
1257 (%compile lambda-expression
1259 :name name
1275 form)
1284 form)
1288 ;;; Handle the evaluation the a :COMPILE-TOPLEVEL body during
1289 ;;; compilation. Normally just evaluate in the appropriate
1290 ;;; environment, but also compile if outputting a CFASL.
1297 ;;; Process a top level FORM with the specified source PATH.
1298 ;;; * If this is a magic top level form, then do stuff.
1299 ;;; * If this is a macro, then expand it.
1300 ;;; * Otherwise, just compile it.
1301 ;;;
1302 ;;; COMPILE-TIME-TOO is as defined in ANSI
1303 ;;; "3.2.3.1 Processing of Top Level Forms".
1314 path)
1319 ;; When we're cross-compiling, consider: what should we
1320 ;; do when we hit e.g.
1321 ;; (EVAL-WHEN (:COMPILE-TOPLEVEL)
1322 ;; (DEFUN FOO (X) (+ 7 X)))?
1323 ;; DEFUN has a macro definition in the cross-compiler,
1324 ;; and a different macro definition in the target
1325 ;; compiler. The only sensible thing is to use the
1326 ;; target compiler's macro definition, since the
1327 ;; cross-compiler's macro is in general into target
1328 ;; functions which can't meaningfully be executed at
1329 ;; cross-compilation time. So make sure we do the EVAL
1330 ;; here, before we macroexpand.
1331 ;;
1332 ;; Then things get even dicier with something like
1333 ;; (DEFCONSTANT-EQX SB!XC:LAMBDA-LIST-KEYWORDS ..)
1334 ;; where we have to make sure that we don't uncross
1335 ;; the SB!XC: prefix before we do EVAL, because otherwise
1336 ;; we'd be trying to redefine the cross-compilation host's
1337 ;; constants.
1338 ;;
1339 ;; (Isn't it fun to cross-compile Common Lisp?:-)
1340 #+sb-xc-host
1345 ;; like SB!XC:DEFCONSTANT and SB!XC:DEFTYPE
1346 ;; should be equivalent to their CL: counterparts
1347 ;; when being compiled as target code. We leave
1348 ;; the rest of the form uncrossed because macros
1349 ;; might yet expand into EVAL-WHEN stuff, and
1350 ;; things inside EVAL-WHEN can't be uncrossed
1351 ;; until after we've EVALed them in the
1352 ;; cross-compilation host.)
1356 slightly-uncrossed)))
1358 ;; (Now that we're no longer processing toplevel
1359 ;; forms, and hence no longer need to worry about
1360 ;; EVAL-WHEN, we can uncross everything.)
1362 ;; (We have to demote COMPILE-TIME-TOO to NIL
1363 ;; here, no matter what it was before, since
1364 ;; otherwise we'd tend to EVAL subforms more than
1365 ;; once, because of WHEN COMPILE-TIME-TOO form
1366 ;; above.)
1368 ;; When we're not cross-compiling, we only need to
1369 ;; macroexpand once, so we can follow the 1-thru-6
1370 ;; sequence of steps in ANSI's "3.2.3.1 Processing of
1371 ;; Top Level Forms".
1372 #-sb-xc-host
1380 path
1381 compile-time-too)))))))
1383 #+sb-xc-host
1384 ;; (There are no xc EVAL-WHEN issues in the ATOM case until
1385 ;; (1) SBCL gets smart enough to handle global
1386 ;; DEFINE-SYMBOL-MACRO or SYMBOL-MACROLET and (2) SBCL
1387 ;; implementors start using symbol macros in a way which
1388 ;; interacts with SB-XC/CL distinction.)
1390 #-sb-xc-host
1396 form))))
1398 ;; In the cross-compiler, top level COLD-FSET arranges
1399 ;; for static linking at cold init time.
1400 #+sb-xc-host
1406 lambda-expression
1407 path)))
1413 ;; CT, LT, and E here are as in Figure 3-7 of ANSI
1414 ;; "3.2.3.1 Processing of Top Level Forms".
1419 e))))
1421 body path new-compile-time-too))
1426 magic
1431 path
1432 compile-time-too))
1436 magic
1440 path
1441 compile-time-too
1451 \f
1452 ;;;; load time value support
1453 ;;;;
1454 ;;;; (See EMIT-MAKE-LOAD-FORM.)
1456 ;;; Return T if we are currently producing a fasl file and hence
1457 ;;; constants need to be dumped carefully.
1461 ;;; Compile FORM and arrange for it to be called at load-time. Return
1462 ;;; the dumper handle and our best guess at the type of the object.
1472 ;;; Compile the FORMS and arrange for them to be called (for effect,
1473 ;;; not value) at load time.
1478 ;;; Do the actual work of COMPILE-LOAD-TIME-VALUE or
1479 ;;; COMPILE-MAKE-LOAD-FORM-INIT-FORMS.
1485 lambda)))
1487 ;;; This is called by COMPILE-TOPLEVEL when it was passed T for
1488 ;;; LOAD-TIME-VALUE-P (which happens in COMPILE-LOAD-TIME-STUFF). We
1489 ;;; don't try to combine this component with anything else and frob
1490 ;;; the name. If not in a :TOPLEVEL component, then don't bother
1491 ;;; compiling, because it was merged with a run-time component.
1500 \f
1501 ;;;; COMPILE-FILE
1511 ;;; Smash LAMBDAS into a single component, compile it, and arrange for
1512 ;;; the resulting function to be called.
1522 ;;; Compile top level code and call the top level lambdas. We pick off
1523 ;;; top level lambdas in non-top-level components here, calling
1524 ;;; SUB-c-t-l-l on each subsequence of normal top level lambdas.
1533 lambdas
1534 ;; this used to read ":start start", but
1535 ;; start can be greater than len, which
1536 ;; is an error according to ANSI - CSR,
1537 ;; 2002-04-25
1539 len)))
1548 ;;; Compile LAMBDAS (a list of CLAMBDAs for top level forms) into the
1549 ;;; object file.
1550 ;;;
1551 ;;; LOAD-TIME-VALUE-P seems to control whether it's MAKE-LOAD-FORM and
1552 ;;; COMPILE-LOAD-TIME-VALUE stuff. -- WHN 20000201
1586 ;;; Actually compile any stuff that has been queued up for block
1587 ;;; compilation.
1607 lexenv))
1611 nil
1627 lexenv))
1637 ;;; Read all forms from INFO and compile them, with output to OBJECT.
1638 ;;; Return (VALUES ABORT-P WARNINGS-P FAILURE-P).
1666 ;; FIXME: Do we need this rebinding here? It's a literal
1667 ;; translation of the old CMU CL rebinding to
1668 ;; (OR *BACKEND-INFO-ENVIRONMENT* *INFO-ENVIRONMENT*),
1669 ;; and it's not obvious whether the rebinding to itself is
1670 ;; needed that SBCL doesn't need *BACKEND-INFO-ENVIRONMENT*.
1683 ;; Dump the code coverage records into the fasl.
1691 list))
1692 nil
1693 nil))
1701 nil)))
1702 ;; Some errors are sufficiently bewildering that we just fail
1703 ;; immediately, without trying to recover and compile more of
1704 ;; the input file.
1710 "~@<~@:_compilation aborted because of fatal error: ~2I~_~A~@:_~:>"
1715 ;;; Return a pathname for the named file. The file must exist.
1724 new
1740 ;;; Print some junk at the beginning and end of compilation.
1748 file-info)
1750 :print-weekday nil
1757 won
1763 ;;; Open some files and call SUB-COMPILE-FILE. If something unwinds
1764 ;;; out of the compile, then abort the writing of the output file, so
1765 ;;; that we don't overwrite it with known garbage.
1768 &key
1770 ;; ANSI options
1772 ;; FIXME: ANSI doesn't seem to say anything about
1773 ;; *COMPILE-VERBOSE* and *COMPILE-PRINT* being rebound by this
1774 ;; function..
1779 ;; extensions
1783 #!+sb-doc
1784 "Compile INPUT-FILE, producing a corresponding fasl file and
1785 returning its filename.
1787 :PRINT
1788 If true, a message per non-macroexpanded top level form is printed
1789 to *STANDARD-OUTPUT*. Top level forms that whose subforms are
1790 processed as top level forms (eg. EVAL-WHEN, MACROLET, PROGN) receive
1791 no such message, but their subforms do.
1793 As an extension to ANSI, if :PRINT is :top-level-forms, a message
1794 per top level form after macroexpansion is printed to *STANDARD-OUTPUT*.
1795 For example, compiling an IN-PACKAGE form will result in a message about
1796 a top level SETQ in addition to the message about the IN-PACKAGE form'
1797 itself.
1799 Both forms of reporting obey the SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*.
1801 :BLOCK-COMPILE
1802 Though COMPILE-FILE accepts an additional :BLOCK-COMPILE
1803 argument, it is not currently supported. (non-standard)
1805 :TRACE-FILE
1806 If given, internal data structures are dumped to the specified
1807 file, or if a value of T is given, to a file of *.trace type
1808 derived from the input file name. (non-standard)
1810 :EMIT-CFASL
1811 (Experimental). If true, outputs the toplevel compile-time effects
1812 of this file into a separate .cfasl file."
1813 ;;; Block compilation is currently broken.
1814 #|
1815 "Also, as a workaround for vaguely-non-ANSI behavior, the
1816 :BLOCK-COMPILE argument is quasi-supported, to determine whether
1817 multiple functions are compiled together as a unit, resolving function
1818 references at compile time. NIL means that global function names are
1819 never resolved at compilation time. Currently NIL is the default
1821 of the ANSI spec allows this behavior under all circumstances, the
1822 compiler's runtime scales badly when it tries to do this for large
1823 files. If/when this performance problem is fixed, the block
1824 compilation default behavior will probably be made dependent on the
1825 SPEED and COMPILATION-SPEED optimization values, and the
1826 :BLOCK-COMPILE argument will probably become deprecated."
1827 |#
1860 default-trace-file-pathname
1862 default-trace-file-pathname))))
1896 ;; CLHS says that the first value is NIL if the "file could not
1897 ;; be created". We interpret this to mean "a valid fasl could not
1898 ;; be created" -- which can happen if the compilation is aborted
1899 ;; before the whole file has been processed, due to eg. a reader
1900 ;; error.
1902 ;; Hack around filesystem race condition...
1904 warnings-p
1905 failure-p)))
1906 \f
1907 ;;; a helper function for COMPILE-FILE-PATHNAME: the default for
1908 ;;; the OUTPUT-FILE argument
1909 ;;;
1910 ;;; ANSI: The defaults for the OUTPUT-FILE are taken from the pathname
1911 ;;; that results from merging the INPUT-FILE with the value of
1912 ;;; *DEFAULT-PATHNAME-DEFAULTS*, except that the type component should
1913 ;;; default to the appropriate implementation-defined default type for
1914 ;;; compiled files.
1918 retyped))
1920 ;;; KLUDGE: Part of the ANSI spec for this seems contradictory:
1921 ;;; If INPUT-FILE is a logical pathname and OUTPUT-FILE is unsupplied,
1922 ;;; the result is a logical pathname. If INPUT-FILE is a logical
1923 ;;; pathname, it is translated into a physical pathname as if by
1924 ;;; calling TRANSLATE-LOGICAL-PATHNAME.
1925 ;;; So I haven't really tried to make this precisely ANSI-compatible
1926 ;;; at the level of e.g. whether it returns logical pathname or a
1927 ;;; physical pathname. Patches to make it more correct are welcome.
1928 ;;; -- WHN 2000-12-09
1930 &key
1933 #!+sb-doc
1934 "Return a pathname describing what file COMPILE-FILE would write to given
1935 these arguments."
1939 \f
1940 ;;;; MAKE-LOAD-FORM stuff
1942 ;;; The entry point for MAKE-LOAD-FORM support. When IR1 conversion
1943 ;;; finds a constant structure, it invokes this to arrange for proper
1944 ;;; dumping. If it turns out that the constant has already been
1945 ;;; dumped, then we don't need to do anything.
1946 ;;;
1947 ;;; If the constant hasn't been dumped, then we check to see whether
1948 ;;; we are in the process of creating it. We detect this by
1949 ;;; maintaining the special *CONSTANTS-BEING-CREATED* as a list of all
1950 ;;; the constants we are in the process of creating. Actually, each
1951 ;;; entry is a list of the constant and any init forms that need to be
1952 ;;; processed on behalf of that constant.
1953 ;;;
1954 ;;; It's not necessarily an error for this to happen. If we are
1955 ;;; processing the init form for some object that showed up *after*
1956 ;;; the original reference to this constant, then we just need to
1957 ;;; defer the processing of that init form. To detect this, we
1958 ;;; maintain *CONSTANTS-CREATED-SINCE-LAST-INIT* as a list of the
1959 ;;; constants created since the last time we started processing an
1960 ;;; init form. If the constant passed to emit-make-load-form shows up
1961 ;;; in this list, then there is a circular chain through creation
1962 ;;; forms, which is an error.
1963 ;;;
1964 ;;; If there is some intervening init form, then we blow out of
1965 ;;; processing it by throwing to the tag PENDING-INIT. The value we
1966 ;;; throw is the entry from *CONSTANTS-BEING-CREATED*. This is so the
1967 ;;; offending init form can be tacked onto the init forms for the
1968 ;;; circular object.
1969 ;;;
1970 ;;; If the constant doesn't show up in *CONSTANTS-BEING-CREATED*, then
1971 ;;; we have to create it. We call MAKE-LOAD-FORM and check to see
1972 ;;; whether the creation form is the magic value
1973 ;;; :SB-JUST-DUMP-IT-NORMALLY. If it is, then we don't do anything. The
1974 ;;; dumper will eventually get its hands on the object and use the
1975 ;;; normal structure dumping noise on it.
1976 ;;;
1977 ;;; Otherwise, we bind *CONSTANTS-BEING-CREATED* and
1978 ;;; *CONSTANTS-CREATED-SINCE- LAST-INIT* and compile the creation form
1979 ;;; much the way LOAD-TIME-VALUE does. When this finishes, we tell the
1980 ;;; dumper to use that result instead whenever it sees this constant.
1981 ;;;
1982 ;;; Now we try to compile the init form. We bind
1983 ;;; *CONSTANTS-CREATED-SINCE-LAST-INIT* to NIL and compile the init
1984 ;;; form (and any init forms that were added because of circularity
1985 ;;; detection). If this works, great. If not, we add the init forms to
1986 ;;; the init forms for the object that caused the problems and let it
1987 ;;; deal with it.
1990 ;;; FIXME: Shouldn't these^ variables be unbound outside LET forms?
1994 ;; KLUDGE: This special hack is because I was too lazy
1995 ;; to rework DEF!STRUCT so that the MAKE-LOAD-FORM
1996 ;; function of LAYOUT returns nontrivial forms when
1997 ;; building the cross-compiler but :IGNORE-IT when
1998 ;; cross-compiling or running under the target Lisp. --
1999 ;; WHN 19990914
2008 ;; If the constant is a reference to a named constant, we can
2009 ;; just use SYMBOL-VALUE during LOAD.
2016 (:sb-just-dump-it-normally
2018 t)
2019 (:ignore-it
2020 nil)
2033 constant
2035 creation-form)
2037 nil)
2039 constant)))
2045 collect name into names
2046 collect form into forms
2048 nil)))
2053 \f
2054 ;;;; Host compile time definitions
2055 #+sb-xc-host
2060 #+sb-xc-host