| blob | 06139e0c8115f97bf29d3169ae6659014a148cf3 |
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?
28 ;;; Whether reference to a thing which cannot be defined causes a full
29 ;;; warning.
34 ;;; Set to NIL to disable loop analysis for register allocation.
37 ;;; Bind this to a stream to capture various internal debugging output.
40 ;;; The current block compilation state. These are initialized to the
41 ;;; :BLOCK-COMPILE and :ENTRY-POINTS arguments that COMPILE-FILE was
42 ;;; called with.
43 ;;;
44 ;;; *BLOCK-COMPILE-ARG* holds the original value of the :BLOCK-COMPILE
45 ;;; argument, which overrides any internal declarations.
52 ;;; When block compiling, used by PROCESS-FORM to accumulate top level
53 ;;; lambdas resulting from compiling subforms. (In reverse order.)
57 ;;; The current non-macroexpanded toplevel form as printed when
58 ;;; *compile-print* is true.
62 #!+sb-doc
65 #!+sb-doc
68 #!+sb-doc
69 "When this is true, the compiler prints to *STANDARD-OUTPUT* progress
70 information about the phases of compilation of each function. (This
74 #!+sb-doc
75 "The defaulted pathname of the file currently being compiled, or NIL if not
78 #!+sb-doc
79 "The TRUENAME of the file currently being compiled, or NIL if not
86 ;;; the SOURCE-INFO structure for the current compilation. This is
87 ;;; null globally to indicate that we aren't currently in any
88 ;;; identifiable compilation.
91 ;;; This is true if we are within a WITH-COMPILATION-UNIT form (which
92 ;;; normally causes nested uses to be no-ops).
95 ;;; Count of the number of compilation units dynamically enclosed by
96 ;;; the current active WITH-COMPILATION-UNIT that were unwound out of.
99 ;;; Mumble conditional on *COMPILE-PROGRESS*.
116 ;; Used during compilation to map code paths to the matching
117 ;; instrumentation conses.
119 ;; Used during compilation to keep track of with source paths have been
120 ;; instrumented in which blocks.
122 ;; Stores the code coverage instrumentation results. Keys are namestrings, the
123 ;; value is a list of (CONS PATH STATE), where STATE is +CODE-COVERAGE-UNMARKED+
124 ;; for a path that has not been visited, and T for one that has.
127 \f
128 ;;;; WITH-COMPILATION-UNIT and WITH-COMPILATION-VALUES
131 #!+sb-doc
132 "Affects compilations that take place within its dynamic extent. It is
133 intended to be eg. wrapped around the compilation of all files in the same system.
135 Following options are defined:
137 :OVERRIDE Boolean-Form
138 One of the effects of this form is to delay undefined warnings until the
139 end of the form, instead of giving them at the end of each compilation.
140 If OVERRIDE is NIL (the default), then the outermost
141 WITH-COMPILATION-UNIT form grabs the undefined warnings. Specifying
142 OVERRIDE true causes that form to grab any enclosed warnings, even if it
143 is enclosed by another WITH-COMPILATION-UNIT.
145 :POLICY Optimize-Declaration-Form
146 Provides dynamic scoping for global compiler optimization qualities and
147 restrictions, limiting effects of subsequent OPTIMIZE proclamations and
148 calls to SB-EXT:RESTRICT-COMPILER-POLICY to the dynamic scope of BODY.
150 If OVERRIDE is false, specified POLICY is merged with current global
151 policy. If OVERRIDE is true, current global policy, including any
152 restrictions, is discarded in favor of the specified POLICY.
154 Supplying POLICY NIL is equivalent to the option not being supplied at
155 all, ie. dynamic scoping of policy does not take place.
157 This option is an SBCL-specific experimental extension: Interface
158 subject to change.
160 :SOURCE-NAMESTRING Namestring-Form
161 Attaches the value returned by the Namestring-Form to the internal
162 debug-source information as the namestring of the source file. Normally
163 the namestring of the input-file for COMPILE-FILE is used: this option
164 can be used to provide source-file information for functions compiled
165 using COMPILE, or to override the input-file of COMPILE-FILE.
167 If both an outer and an inner WITH-COMPILATION-UNIT provide a
168 SOURCE-NAMESTRING, the inner one takes precedence. Unaffected
169 by :OVERRIDE.
171 This is an SBCL-specific extension.
173 :SOURCE-PLIST Plist-Form
174 Attaches the value returned by the Plist-Form to internal debug-source
175 information of functions compiled in within the dynamic extent of BODY.
177 Primarily for use by development environments, in order to eg. associate
178 function definitions with editor-buffers. Can be accessed using
179 SB-INTROSPECT:DEFINITION-SOURCE-PLIST.
181 If an outer WITH-COMPILATION-UNIT form also provide a SOURCE-PLIST, it
182 is appended to the end of the provided SOURCE-PLIST. Unaffected
183 by :OVERRIDE.
185 This is an SBCL-specific extension.
187 Examples:
189 ;; Prevent proclamations from the file leaking, and restrict
190 ;; SAFETY to 3 -- otherwise uses the current global policy.
191 (with-compilation-unit (:policy '(optimize))
192 (restrict-compiler-policy 'safety 3)
195 ;; Using default policy instead of the current global one,
196 ;; except for DEBUG 3.
197 (with-compilation-unit (:policy '(optimize debug)
198 :override t)
201 ;; Same as if :POLICY had not been specified at all: SAFETY 3
202 ;; proclamation leaks out from WITH-COMPILATION-UNIT.
203 (with-compilation-unit (:policy nil)
204 (declaim (optimize safety))
206 "
221 ;; Inside another WITH-COMPILATION-UNIT, a WITH-COMPILATION-UNIT is
222 ;; ordinarily (unless OVERRIDE) basically a no-op.
250 ;;; Is NAME something that no conforming program can rely on
251 ;;; defining?
254 (:function
257 (:type
264 ;;; This is to be called at the end of a compilation unit. It signals
265 ;;; any residual warnings about unknown stuff, then prints the total
266 ;;; error counts. ABORT-P should be true when the compilation unit was
267 ;;; aborted by throwing out. ABORT-COUNT is the number of dynamically
268 ;;; enclosed nested compilation units that were aborted.
295 #+sb-xc-host nil
297 (:function
299 "~@<The function ~S is undefined, and its name is ~
300 reserved by ANSI CL so that even if it were ~
301 defined later, the code doing so would not be ~
303 (:type
306 "~@<Undefined type ~S. The name starts with ~S: ~
307 probably use of a quoted type name in a context ~
308 where the name is not evaluated.~:@>"
311 "~@<Undefined type ~S. Note that name ~S is ~
312 reserved by ANSI CL, so code defining a type with ~
313 that name would not be portable.~:@>" name
314 name))))
323 "~W more use~:P of undefined ~(~A~) ~S"
324 more kind name)
326 "~W more use~:P of undefined ~(~A~) ~S"
327 more kind name))))))))))
337 abort-p)
341 "~& Undefined ~(~A~)~p:~
342 ~% ~{~<~% ~1:;~S~>~^ ~}"
345 ~[~:;~:*~& caught ~W ERROR condition~:P~]~
346 ~[~:;~:*~& caught ~W WARNING condition~:P~]~
347 ~[~:;~:*~& caught ~W STYLE-WARNING condition~:P~]~
348 ~[~:;~:*~& printed ~W note~:P~]"
357 ;; Bidrectional map between IR1/IR2/assembler abstractions
358 ;; and a corresponding small integer identifier. One direction could be done
359 ;; by adding the integer ID as an object slot, but we want both directions.
375 ;;; Evaluate BODY, then return (VALUES BODY-VALUE WARNINGS-P
376 ;;; FAILURE-P), where BODY-VALUE is the first value of the body, and
377 ;;; WARNINGS-P and FAILURE-P are as in CL:COMPILE or CL:COMPILE-FILE.
379 ;; This binding could just as well be in WITH-IR1-NAMESPACE, but
380 ;; since it's primarily a debugging tool, it's nicer to have
381 ;; a wider unique scope by ID.
396 ;;; THING is a kind of thing about which we'd like to issue a warning,
397 ;;; but showing at most one warning for a given set of <THING,FMT,ARGS>.
398 ;;; The compiler does a good job of making sure not to print repetitive
399 ;;; warnings for code that it compiles, but this solves a different problem.
400 ;;; Specifically, for a warning from PARSE-LAMBDA-LIST, there are three calls:
401 ;;; - once in the expander for defmacro itself, as it calls MAKE-MACRO-LAMBDA
402 ;;; which calls PARSE-LAMBDA-LIST. This is the toplevel form processing.
403 ;;; - again for :compile-toplevel, where the DS-BIND calls PARSE-LAMBDA-LIST.
404 ;;; If compiling in compile-toplevel, then *COMPILE-OBJECT* is a core object,
405 ;;; but if interpreting, then it is still a fasl.
406 ;;; - once for compiling to fasl. *COMPILE-OBJECT* is a fasl.
407 ;;; I'd have liked the data to be associated with the fasl, except that
408 ;;; as indicated above, the second line hides some information.
416 ;; THING is compared by EQ, FMT by STRING=.
419 ;; We don't want to walk into default values,
420 ;; e.g. (&optional (b #<insane-struct))
421 ;; because #<insane-struct> might be circular.
430 \f
431 ;;;; component compilation
434 #!+sb-doc
435 "The upper limit on the number of times that we will consecutively do IR1
436 optimization that doesn't introduce any new code. A finite limit is
442 ;;; Repeatedly optimize COMPONENT until no further optimizations can
443 ;;; be found or we hit our iteration limit. When we hit the limit, we
444 ;;; clear the component and block REOPTIMIZE flags to discourage the
445 ;;; next optimization attempt from pounding on the same code.
489 ;;; KLUDGE: This was bumped from 5 to 10 in a DTC patch ported by MNA
490 ;;; from CMU CL into sbcl-0.6.11.44, the same one which allowed IR1
491 ;;; transforms to be delayed. Either DTC or MNA or both didn't explain
492 ;;; why, and I don't know what the rationale was. -- WHN 2001-04-28
493 ;;;
494 ;;; FIXME: It would be good to document why it's important to have a
495 ;;; large value here, and what the drawbacks of an excessively large
496 ;;; value are; and it might also be good to make it depend on
497 ;;; optimization policy.
503 ;;; Iterate doing FIND-DFO until no new dead code is discovered.
516 ;;; Do all the IR1 phases for a non-top-level component.
521 ;make this 0.
544 ;; We delay the generation of type checks until
545 ;; the type constraints have had time to
546 ;; propagate, else the compiler can confuse itself.
585 ;; Assign BLOCK-NUMBER for any cleanup blocks introduced by
586 ;; stack analysis. There shouldn't be any unreachable code after
587 ;; control, so this won't delete anything.
642 #-sb-xc-host
654 code-length
655 fixup-notes
662 code-length
663 fixup-notes
667 ;; We're done, so don't bother keeping anything around.
672 ;;; Delete components with no external entry points before we try to
673 ;;; generate code. Unreachable closures can cause IR2 conversion to
674 ;;; puke on itself, since it is the reference to the closure which
675 ;;; normally causes the components to be combined.
682 (:external
690 ;; miscellaneous sanity checks
691 ;;
692 ;; FIXME: These are basically pretty wimpy compared to the checks done
693 ;; by the old CHECK-IR1-CONSISTENCY code. It would be really nice to
694 ;; make those internal consistency checks work again and use them.
699 ;; sanity check to prevent weirdness from propagating insidiously as
700 ;; far from its root cause as it did in bug 138: Make sure that
701 ;; thing-to-COMPONENT links are consistent.
707 ;; Record xref information before optimization. This way the
708 ;; stored xref data reflects the real source as closely as
709 ;; possible.
719 #|
731 |#
733 ;; This should happen at some point before PHYSENV-ANALYZE, and
734 ;; after RECORD-COMPONENT-XREFS. Beyond that, I haven't really
735 ;; thought things through. -- AJB, 2014-Jun-08
738 ;; FIXME: What is MAYBE-MUMBLE for? Do we need it any more?
752 \f
753 ;;;; clearing global data structures
754 ;;;;
755 ;;;; FIXME: Is it possible to get rid of this stuff, getting rid of
756 ;;;; global data structures entirely when possible and consing up the
757 ;;;; others from scratch instead of clearing and reusing them?
759 ;;; Clear the INFO in constants in the *FREE-VARS*, etc. In
760 ;;; addition to allowing stuff to be reclaimed, this is required for
761 ;;; correct assignment of constant offsets, since we need to assign a
762 ;;; new offset for each component. We don't clear the FUNCTIONAL-INFO
763 ;;; slots, since they are used to keep track of functions across
764 ;;; component boundaries.
779 ;;; Blow away the REFS for all global variables, and let COMPONENT
780 ;;; be recycled.
792 x))
799 \f
800 ;;;; trace output
802 ;;; Print out some useful info about COMPONENT to STREAM.
823 \f
824 ;;; Given a pathname, return a SOURCE-INFO structure.
829 :external-format external-format
830 :subforms
835 ;;; Return a SOURCE-INFO to describe the incremental compilation of FORM.
843 ;;; Walk up the SOURCE-INFO list until we either reach a SOURCE-INFO
844 ;;; with no parent (e.g., from a REPL evaluation) or until we reach a
845 ;;; SOURCE-INFO whose FILE-INFO denotes a file.
853 finfo))))
855 ;;; If STREAM is present, return it, otherwise open a stream to the
856 ;;; current file. There must be a current file.
857 ;;;
858 ;;; FIXME: This is probably an unnecessarily roundabout way to do
859 ;;; things now that we process a single file in COMPILE-FILE (unlike
860 ;;; the old CMU CL code, which accepted multiple files). Also, the old
861 ;;; comment said
862 ;;; When we open a new file, we also reset *PACKAGE* and policy.
863 ;;; This gives the effect of rebinding around each file.
864 ;;; which doesn't seem to be true now. Check to make sure that if
865 ;;; such rebinding is necessary, it's still done somewhere.
878 :external-format external-format
879 ;; SBCL stream classes aren't available in the host
885 stream)))))
887 ;;; Close the stream in INFO if it is open.
895 ;; Loop over forms read from INFO's stream, calling FUNCTION with each.
896 ;; CONDITION-NAME is signaled if there is a reader error, and should be
897 ;; a subtype of not-so-aptly-named INPUT-ERROR-IN-COMPILE-FILE.
904 ;; Return a form read from STREAM; or for EOF use the trick,
905 ;; popularized by Kent Pitman, of returning STREAM itself.
908 ;; Reset for a new toplevel form.
916 ;; We don't need to supply :POSITION here because
917 ;; READER-ERRORs already know their position in the file.
918 :condition condition
920 ;; ANSI, in its wisdom, says that READ should return END-OF-FILE
921 ;; (and that this is not a READER-ERROR) when it encounters end of
922 ;; file in the middle of something it's trying to read,
923 ;; making it unfortunately indistinguishable from legal EOF.
924 ;; Were it not for that, it would be more elegant to just
925 ;; handle one more condition in the HANDLER-CASE.
928 condition-name
929 :condition condition
930 ;; We need to supply :POSITION here because the END-OF-FILE
931 ;; condition doesn't carry the position that the user
932 ;; probably cares about, where the failed READ began.
933 :position
936 pos)
940 stream
945 :current-index
950 current-idx))
953 ;;; Loop over FORMS retrieved from INFO. Used by COMPILE-FILE and
954 ;;; LOAD when loading from a FILE-STREAM associated with a source
955 ;;; file. ON-ERROR is the name of a condition class that should
956 ;;; be signaled if anything goes wrong during a READ.
967 ;;; Read and compile the source file.
975 ;; It's easy to get into a situation where cold-init crashes and the only
976 ;; backtrace you get from ldb is TOP-LEVEL-FORM, which means you're anywhere
977 ;; within the 23000 or so blobs of code deferred until cold-init.
978 ;; Seeing each file finish narrows things down without the noise of :sb-show,
979 ;; but this hack messes up form positions, so it's not on unless asked for.
993 ;;; Return the INDEX'th source form read from INFO and the position
994 ;;; where it was read.
1000 \f
1001 ;;;; processing of top level forms
1003 ;;; This is called by top level form processing when we are ready to
1004 ;;; actually compile something. If *BLOCK-COMPILE* is T, then we still
1005 ;;; convert the form, but delay compilation, pushing the result on
1006 ;;; *TOPLEVEL-LAMBDAS* instead.
1009 #+sb-xc-host
1013 ;; Don't bother to compile simple objects that just sit there.
1027 nil))))))
1029 ;;; Macroexpand FORM in the current environment with an error handler.
1030 ;;; We only expand one level, so that we retain all the intervening
1031 ;;; forms in the source path. A compiler-macro takes precedence over
1032 ;;; an ordinary macro as specified in CLHS 3.2.3.1
1033 ;;; Note that this function is _only_ for processing of toplevel forms.
1034 ;;; Non-toplevel forms use IR1-CONVERT-FUNCTOID which considers compiler macros.
1047 condition))))
1049 ;;; Process a PROGN-like portion of a top level form. FORMS is a list of
1050 ;;; the forms, and PATH is the source path of the FORM they came out of.
1051 ;;; COMPILE-TIME-TOO is as in ANSI "3.2.3.1 Processing of Top Level Forms".
1057 ;;; Process a top level use of LOCALLY, or anything else (e.g.
1058 ;;; MACROLET) at top level which has declarations and ordinary forms.
1059 ;;; We parse declarations and then recursively process the body.
1065 ;; FIXME: VALUES declaration
1066 ;;
1067 ;; Binding *POLICY* is pretty much of a hack, since it
1068 ;; causes LOCALLY to "capture" enclosed proclamations. It
1069 ;; is necessary because CONVERT-AND-MAYBE-COMPILE uses the
1070 ;; value of *POLICY* as the policy. The need for this hack
1071 ;; is due to the quirk that there is no way to represent in
1072 ;; a POLICY that an optimize quality came from the default.
1073 ;;
1074 ;; FIXME: Ideally, something should be done so that DECLAIM
1075 ;; inside LOCALLY works OK. Failing that, at least we could
1076 ;; issue a warning instead of silently screwing up.
1077 ;; Here's how to fix this: a POLICY object can in fact represent
1078 ;; absence of qualitities. Whenever we rebind *POLICY* (here and
1079 ;; elsewhere), it should be bound to a policy that expresses no
1080 ;; qualities. Proclamations should update SYMBOL-GLOBAL-VALUE of
1081 ;; *POLICY*, which can be seen irrespective of dynamic bindings,
1082 ;; and declarations should update the lexical policy.
1083 ;; The POLICY macro can be amended to merge the dynamic *POLICY*
1084 ;; (or whatever it came from, like a LEXENV) with the global
1085 ;; *POLICY*. COERCE-TO-POLICY can do the merge, employing a 1-line
1086 ;; cache so that repeated calls for any two fixed policy objects
1087 ;; return the identical value (since policies are immutable).
1089 ;; This is probably also a hack
1091 ;; ditto
1095 ;;; Parse an EVAL-WHEN situations list, returning three flags,
1096 ;;; (VALUES COMPILE-TOPLEVEL LOAD-TOPLEVEL EXECUTE), indicating
1097 ;;; the types of situations present in the list.
1101 '(:compile-toplevel
1102 compile
1103 :load-toplevel
1104 load
1105 :execute
1106 eval)))
1111 deprecated-names)))
1113 situations)
1118 ;;; utilities for extracting COMPONENTs of FUNCTIONALs
1129 result))))
1133 result))))
1136 &key
1137 name
1139 ;; I'd thought NIL should
1140 ;; work, but it doesn't.
1141 ;; -- WHN 2001-09-20
1152 lambda-expression
1154 ;; Convert the XEP using the policy of the real function. Otherwise
1155 ;; the wrong policy will be used for deciding whether to type-check
1156 ;; the parameters of the real function (via CONVERT-CALL /
1157 ;; PROPAGATE-TO-ARGS). -- JES, 2007-02-27
1160 :source-name source-name
1172 ;; Any leftover REFs to FUN outside local calls get replaced with the
1173 ;; XEP.
1180 xep
1181 fun)
1182 xep)))
1184 ;;; Compile LAMBDA-EXPRESSION into *COMPILE-OBJECT*, returning a
1185 ;;; description of the result.
1186 ;;; * If *COMPILE-OBJECT* is a CORE-OBJECT, then write the function
1187 ;;; into core and return the compiled FUNCTION value.
1188 ;;; * If *COMPILE-OBJECT* is a fasl file, then write the function
1189 ;;; into the fasl file and return a dump handle.
1190 ;;;
1191 ;;; If NAME is provided, then we try to use it as the name of the
1192 ;;; function for debugging/diagnostic information.
1195 &key
1196 name
1198 ;; This magical idiom seems to be the appropriate
1199 ;; path for compiling standalone LAMBDAs, judging
1200 ;; from the CMU CL code and experiment, so it's a
1201 ;; nice default for things where we don't have a
1202 ;; real source path (as in e.g. inside CL:COMPILE).
1213 :name name
1216 ;; FIXME: The compile-it code from here on is sort of a
1217 ;; twisted version of the code in COMPILE-TOPLEVEL. It'd be
1218 ;; better to find a way to share the code there; or
1219 ;; alternatively, to use this code to replace the code there.
1220 ;; (The second alternative might be pretty easy if we used
1221 ;; the :LOCALL-ONLY option to IR1-FOR-LAMBDA. Then maybe the
1222 ;; whole FUNCTIONAL-KIND=:TOPLEVEL case could go away..)
1240 result
1241 ;; KLUDGE: This code duplicates some other code in this
1242 ;; file. In the great reorganzation, the flow of program
1243 ;; logic changed from the original CMUCL model, and that
1244 ;; path (as of sbcl-0.7.5 in SUB-COMPILE-FILE) was no
1245 ;; longer followed for CORE-OBJECTS, leading to BUG
1246 ;; 156. This place is transparently not the right one for
1247 ;; this code, but I don't have a clear enough overview of
1248 ;; the compiler to know how to rearrange it all so that
1249 ;; this operation fits in nicely, and it was blocking
1250 ;; reimplementation of (DECLAIM (INLINE FOO)) (MACROLET
1251 ;; ((..)) (DEFUN FOO ...))
1252 ;;
1253 ;; FIXME: This KLUDGE doesn't solve all the problem in an
1254 ;; ideal way, as (1) definitions typed in at the REPL
1255 ;; without an INLINE declaration will give a NULL
1256 ;; FUNCTION-LAMBDA-EXPRESSION (allowable, but not ideal)
1257 ;; and (2) INLINE declarations will yield a
1258 ;; FUNCTION-LAMBDA-EXPRESSION headed by
1259 ;; SB-C:LAMBDA-WITH-LEXENV, even for null LEXENV. -- CSR,
1260 ;; 2002-07-02
1261 ;;
1262 ;; (2) is probably fairly easy to fix -- it is, after all,
1263 ;; a matter of list manipulation (or possibly of teaching
1264 ;; CL:FUNCTION about SB-C:LAMBDA-WITH-LEXENV). (1) is
1265 ;; significantly harder, as the association between
1266 ;; function object and source is a tricky one.
1267 ;;
1268 ;; FUNCTION-LAMBDA-EXPRESSION "works" (i.e. returns a
1269 ;; non-NULL list) when the function in question has been
1270 ;; compiled by (COMPILE <x> '(LAMBDA ...)); it does not
1271 ;; work when it has been compiled as part of the top-level
1272 ;; EVAL strategy of compiling everything inside (LAMBDA ()
1273 ;; ...). -- CSR, 2002-11-02
1290 form)
1299 form)
1303 ;;; Handle the evaluation the a :COMPILE-TOPLEVEL body during
1304 ;;; compilation. Normally just evaluate in the appropriate
1305 ;;; environment, but also compile if outputting a CFASL.
1320 ;; In case a null lexenv is created, it needs to get the newly
1321 ;; effective global policy, not the policy currently in *POLICY*.
1325 ;;; Process a top level FORM with the specified source PATH.
1326 ;;; * If this is a magic top level form, then do stuff.
1327 ;;; * If this is a macro, then expand it.
1328 ;;; * Otherwise, just compile it.
1329 ;;;
1330 ;;; COMPILE-TIME-TOO is as defined in ANSI
1331 ;;; "3.2.3.1 Processing of Top Level Forms".
1342 path)
1347 ;; When we're cross-compiling, consider: what should we
1348 ;; do when we hit e.g.
1349 ;; (EVAL-WHEN (:COMPILE-TOPLEVEL)
1350 ;; (DEFUN FOO (X) (+ 7 X)))?
1351 ;; DEFUN has a macro definition in the cross-compiler,
1352 ;; and a different macro definition in the target
1353 ;; compiler. The only sensible thing is to use the
1354 ;; target compiler's macro definition, since the
1355 ;; cross-compiler's macro is in general into target
1356 ;; functions which can't meaningfully be executed at
1357 ;; cross-compilation time. So make sure we do the EVAL
1358 ;; here, before we macroexpand.
1359 ;;
1360 ;; Then things get even dicier with something like
1361 ;; (DEFCONSTANT-EQX SB!XC:LAMBDA-LIST-KEYWORDS ..)
1362 ;; where we have to make sure that we don't uncross
1363 ;; the SB!XC: prefix before we do EVAL, because otherwise
1364 ;; we'd be trying to redefine the cross-compilation host's
1365 ;; constants.
1366 ;;
1367 ;; (Isn't it fun to cross-compile Common Lisp?:-)
1368 #+sb-xc-host
1373 ;; like SB!XC:DEFCONSTANT and SB!XC:DEFTYPE
1374 ;; should be equivalent to their CL: counterparts
1375 ;; when being compiled as target code. We leave
1376 ;; the rest of the form uncrossed because macros
1377 ;; might yet expand into EVAL-WHEN stuff, and
1378 ;; things inside EVAL-WHEN can't be uncrossed
1379 ;; until after we've EVALed them in the
1380 ;; cross-compilation host.)
1384 slightly-uncrossed)))
1386 ;; (Now that we're no longer processing toplevel
1387 ;; forms, and hence no longer need to worry about
1388 ;; EVAL-WHEN, we can uncross everything.)
1390 ;; (We have to demote COMPILE-TIME-TOO to NIL
1391 ;; here, no matter what it was before, since
1392 ;; otherwise we'd tend to EVAL subforms more than
1393 ;; once, because of WHEN COMPILE-TIME-TOO form
1394 ;; above.)
1396 ;; When we're not cross-compiling, we only need to
1397 ;; macroexpand once, so we can follow the 1-thru-6
1398 ;; sequence of steps in ANSI's "3.2.3.1 Processing of
1399 ;; Top Level Forms".
1400 #-sb-xc-host
1408 path
1409 compile-time-too)))))))
1411 #+sb-xc-host
1412 ;; (There are no xc EVAL-WHEN issues in the ATOM case until
1413 ;; (1) SBCL gets smart enough to handle global
1414 ;; DEFINE-SYMBOL-MACRO or SYMBOL-MACROLET and (2) SBCL
1415 ;; implementors start using symbol macros in a way which
1416 ;; interacts with SB-XC/CL distinction.)
1418 #-sb-xc-host
1424 form))))
1431 ;; CT, LT, and E here are as in Figure 3-7 of ANSI
1432 ;; "3.2.3.1 Processing of Top Level Forms".
1437 e))))
1439 body path new-compile-time-too))
1444 magic
1449 path
1450 compile-time-too))
1454 magic
1458 path
1459 compile-time-too
1469 \f
1470 ;;;; load time value support
1471 ;;;;
1472 ;;;; (See EMIT-MAKE-LOAD-FORM.)
1474 ;;; Return T if we are currently producing a fasl file and hence
1475 ;;; constants need to be dumped carefully.
1480 ;;; Compile the FORMS and arrange for them to be called (for effect,
1481 ;;; not value) at load time.
1483 ;; If FORMS has exactly one PROGN containing a call of SB-PCL::SET-SLOTS,
1484 ;; then fopcompile it, otherwise use the main compiler.
1497 ;; invoke recursive MAKE-LOAD-FORM stuff as necessary
1508 ;;; Do the actual work of COMPILE-LOAD-TIME-VALUE or
1509 ;;; COMPILE-MAKE-LOAD-FORM-INIT-FORMS.
1515 lambda)))
1517 ;;; This is called by COMPILE-TOPLEVEL when it was passed T for
1518 ;;; LOAD-TIME-VALUE-P (which happens in COMPILE-LOAD-TIME-STUFF). We
1519 ;;; don't try to combine this component with anything else and frob
1520 ;;; the name. If not in a :TOPLEVEL component, then don't bother
1521 ;;; compiling, because it was merged with a run-time component.
1530 \f
1531 ;;;; COMPILE-FILE
1541 ;;; Smash LAMBDAS into a single component, compile it, and arrange for
1542 ;;; the resulting function to be called.
1552 ;;; Compile top level code and call the top level lambdas. We pick off
1553 ;;; top level lambdas in non-top-level components here, calling
1554 ;;; SUB-c-t-l-l on each subsequence of normal top level lambdas.
1563 lambdas
1564 ;; this used to read ":start start", but
1565 ;; start can be greater than len, which
1566 ;; is an error according to ANSI - CSR,
1567 ;; 2002-04-25
1569 len)))
1578 ;;; Compile LAMBDAS (a list of CLAMBDAs for top level forms) into the
1579 ;;; object file.
1580 ;;;
1581 ;;; LOAD-TIME-VALUE-P seems to control whether it's MAKE-LOAD-FORM and
1582 ;;; COMPILE-LOAD-TIME-VALUE stuff. -- WHN 20000201
1615 ;;; Actually compile any stuff that has been queued up for block
1616 ;;; compilation.
1635 lexenv))
1636 ;; Is this right? I would think that if lexenv is null
1637 ;; we should look at *HANDLED-CONDITIONS*.
1660 ;; WOULD-MUFFLE-P is called (incorrectly) only by NOTE-UNDEFINED-REFERENCE.
1661 ;; It is not wrong per se, but as used, it is wrong, making it nearly
1662 ;; impossible to muffle a subset of undefind warnings whose NAME and KIND
1663 ;; slots match specific things tested by a user-defined predicate.
1664 ;; Attempting to do that might muffle everything, depending on how your
1665 ;; predicate responds to a vanilla WARNING. Consider e.g.
1666 ;; (AND WARNING (NOT (SATISFIES HAIRYFN)))
1667 ;; where HAIRYFN depends on the :FORMAT-CONTROL and :FORMAT-ARGUMENTS.
1673 ;;; Read all forms from INFO and compile them, with output to OBJECT.
1674 ;;; Return (VALUES ABORT-P WARNINGS-P FAILURE-P).
1709 ;; Dump the code coverage records into the fasl.
1718 list))
1719 nil
1720 nil)))
1727 nil))))
1728 ;; Some errors are sufficiently bewildering that we just fail
1729 ;; immediately, without trying to recover and compile more of
1730 ;; the input file.
1736 "~@<~@:_compilation aborted because of fatal error: ~2I~_~A~@:_~:>"
1741 ;;; Return a pathname for the named file. The file must exist.
1750 new
1766 ;;; Print some junk at the beginning and end of compilation.
1772 "preloading"
1778 file-info)
1780 :print-weekday nil
1787 won
1793 ;;; Open some files and call SUB-COMPILE-FILE. If something unwinds
1794 ;;; out of the compile, then abort the writing of the output file, so
1795 ;;; that we don't overwrite it with known garbage.
1798 &key
1800 ;; ANSI options
1802 ;; FIXME: ANSI doesn't seem to say anything about
1803 ;; *COMPILE-VERBOSE* and *COMPILE-PRINT* being rebound by this
1804 ;; function..
1809 ;; extensions
1813 #!+sb-doc
1814 "Compile INPUT-FILE, producing a corresponding fasl file and
1815 returning its filename.
1817 :PRINT
1818 If true, a message per non-macroexpanded top level form is printed
1819 to *STANDARD-OUTPUT*. Top level forms that whose subforms are
1820 processed as top level forms (eg. EVAL-WHEN, MACROLET, PROGN) receive
1821 no such message, but their subforms do.
1823 As an extension to ANSI, if :PRINT is :top-level-forms, a message
1824 per top level form after macroexpansion is printed to *STANDARD-OUTPUT*.
1825 For example, compiling an IN-PACKAGE form will result in a message about
1826 a top level SETQ in addition to the message about the IN-PACKAGE form'
1827 itself.
1829 Both forms of reporting obey the SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*.
1831 :BLOCK-COMPILE
1832 Though COMPILE-FILE accepts an additional :BLOCK-COMPILE
1833 argument, it is not currently supported. (non-standard)
1835 :TRACE-FILE
1836 If given, internal data structures are dumped to the specified
1837 file, or if a value of T is given, to a file of *.trace type
1838 derived from the input file name. (non-standard)
1840 :EMIT-CFASL
1841 (Experimental). If true, outputs the toplevel compile-time effects
1842 of this file into a separate .cfasl file."
1843 ;;; Block compilation is currently broken.
1844 #|
1845 "Also, as a workaround for vaguely-non-ANSI behavior, the
1846 :BLOCK-COMPILE argument is quasi-supported, to determine whether
1847 multiple functions are compiled together as a unit, resolving function
1848 references at compile time. NIL means that global function names are
1849 never resolved at compilation time. Currently NIL is the default
1851 of the ANSI spec allows this behavior under all circumstances, the
1852 compiler's runtime scales badly when it tries to do this for large
1853 files. If/when this performance problem is fixed, the block
1854 compilation default behavior will probably be made dependent on the
1855 SPEED and COMPILATION-SPEED optimization values, and the
1856 :BLOCK-COMPILE argument will probably become deprecated."
1857 |#
1892 default-trace-file-pathname
1894 default-trace-file-pathname))))
1928 ;; CLHS says that the first value is NIL if the "file could not
1929 ;; be created". We interpret this to mean "a valid fasl could not
1930 ;; be created" -- which can happen if the compilation is aborted
1931 ;; before the whole file has been processed, due to eg. a reader
1932 ;; error.
1934 ;; Hack around filesystem race condition...
1936 warnings-p
1937 failure-p)))
1938 \f
1939 ;;; a helper function for COMPILE-FILE-PATHNAME: the default for
1940 ;;; the OUTPUT-FILE argument
1941 ;;;
1942 ;;; ANSI: The defaults for the OUTPUT-FILE are taken from the pathname
1943 ;;; that results from merging the INPUT-FILE with the value of
1944 ;;; *DEFAULT-PATHNAME-DEFAULTS*, except that the type component should
1945 ;;; default to the appropriate implementation-defined default type for
1946 ;;; compiled files.
1950 retyped))
1952 ;;; KLUDGE: Part of the ANSI spec for this seems contradictory:
1953 ;;; If INPUT-FILE is a logical pathname and OUTPUT-FILE is unsupplied,
1954 ;;; the result is a logical pathname. If INPUT-FILE is a logical
1955 ;;; pathname, it is translated into a physical pathname as if by
1956 ;;; calling TRANSLATE-LOGICAL-PATHNAME.
1957 ;;; So I haven't really tried to make this precisely ANSI-compatible
1958 ;;; at the level of e.g. whether it returns logical pathname or a
1959 ;;; physical pathname. Patches to make it more correct are welcome.
1960 ;;; -- WHN 2000-12-09
1962 &key
1965 #!+sb-doc
1966 "Return a pathname describing what file COMPILE-FILE would write to given
1967 these arguments."
1971 \f
1972 ;;;; MAKE-LOAD-FORM stuff
1974 ;;; The entry point for MAKE-LOAD-FORM support. When IR1 conversion
1975 ;;; finds a constant structure, it invokes this to arrange for proper
1976 ;;; dumping. If it turns out that the constant has already been
1977 ;;; dumped, then we don't need to do anything.
1978 ;;;
1979 ;;; If the constant hasn't been dumped, then we check to see whether
1980 ;;; we are in the process of creating it. We detect this by
1981 ;;; maintaining the special *CONSTANTS-BEING-CREATED* as a list of all
1982 ;;; the constants we are in the process of creating. Actually, each
1983 ;;; entry is a list of the constant and any init forms that need to be
1984 ;;; processed on behalf of that constant.
1985 ;;;
1986 ;;; It's not necessarily an error for this to happen. If we are
1987 ;;; processing the init form for some object that showed up *after*
1988 ;;; the original reference to this constant, then we just need to
1989 ;;; defer the processing of that init form. To detect this, we
1990 ;;; maintain *CONSTANTS-CREATED-SINCE-LAST-INIT* as a list of the
1991 ;;; constants created since the last time we started processing an
1992 ;;; init form. If the constant passed to emit-make-load-form shows up
1993 ;;; in this list, then there is a circular chain through creation
1994 ;;; forms, which is an error.
1995 ;;;
1996 ;;; If there is some intervening init form, then we blow out of
1997 ;;; processing it by throwing to the tag PENDING-INIT. The value we
1998 ;;; throw is the entry from *CONSTANTS-BEING-CREATED*. This is so the
1999 ;;; offending init form can be tacked onto the init forms for the
2000 ;;; circular object.
2001 ;;;
2002 ;;; If the constant doesn't show up in *CONSTANTS-BEING-CREATED*, then
2003 ;;; we have to create it. We call %MAKE-LOAD-FORM and check
2004 ;;; if the result is 'FOP-STRUCT, and if so we don't do anything.
2005 ;;; The dumper will eventually get its hands on the object and use the
2006 ;;; normal structure dumping noise on it.
2007 ;;;
2008 ;;; Otherwise, we bind *CONSTANTS-BEING-CREATED* and
2009 ;;; *CONSTANTS-CREATED-SINCE- LAST-INIT* and compile the creation form
2010 ;;; much the way LOAD-TIME-VALUE does. When this finishes, we tell the
2011 ;;; dumper to use that result instead whenever it sees this constant.
2012 ;;;
2013 ;;; Now we try to compile the init form. We bind
2014 ;;; *CONSTANTS-CREATED-SINCE-LAST-INIT* to NIL and compile the init
2015 ;;; form (and any init forms that were added because of circularity
2016 ;;; detection). If this works, great. If not, we add the init forms to
2017 ;;; the init forms for the object that caused the problems and let it
2018 ;;; deal with it.
2021 ;;; FIXME: Shouldn't these^ variables be unbound outside LET forms?
2031 ;; If this is a global constant reference, we can call SYMBOL-GLOBAL-VALUE
2032 ;; during LOAD as a fasl op, and not compile a lambda.
2041 t)
2042 (:ignore-it
2043 nil)
2056 constant
2064 index))
2067 fasl)
2068 nil)
2070 constant)))
2076 collect form into forms
2078 nil)))
2082 nil)))))
2084 \f
2085 ;;;; Host compile time definitions
2086 #+sb-xc-host
2091 #+sb-xc-host