| blob | 2c413654ba3949ad8f52b3320adc34b2df3b25fb |
1 ;;;; This file implements local call analysis. A local call is a
2 ;;;; function call between functions being compiled at the same time.
3 ;;;; If we can tell at compile time that such a call is legal, then we
4 ;;;; change the combination to call the correct lambda, mark it as
5 ;;;; local, and add this link to our call graph. Once a call is local,
6 ;;;; it is then eligible for let conversion, which places the body of
7 ;;;; the function inline.
8 ;;;;
9 ;;;; We cannot always do a local call even when we do have the
10 ;;;; function being called. Calls that cannot be shown to have legal
11 ;;;; arg counts are not converted.
13 ;;;; This software is part of the SBCL system. See the README file for
14 ;;;; more information.
15 ;;;;
16 ;;;; This software is derived from the CMU CL system, which was
17 ;;;; written at Carnegie Mellon University and released into the
18 ;;;; public domain. The software is in the public domain and is
19 ;;;; provided with absolutely no warranty. See the COPYING and CREDITS
20 ;;;; files for more information.
24 ;;; This function propagates information from the variables in the
25 ;;; function FUN to the actual arguments in CALL. This is also called
26 ;;; by the VALUES IR1 optimizer when it sleazily converts MV-BINDs to
27 ;;; LETs.
28 ;;;
29 ;;; We flush all arguments to CALL that correspond to unreferenced
30 ;;; variables in FUN. We leave NILs in the COMBINATION-ARGS so that
31 ;;; the remaining args still match up with their vars.
32 ;;;
33 ;;; We also apply the declared variable type assertion to the argument
34 ;;; lvars.
53 collect arg into dx-lvars
55 ;; Stack analysis wants DX value generators to end
56 ;; their blocks. Uses of mupltiple used LVARs already
57 ;; end their blocks, so we just need to process
58 ;; used-once LVARs.
66 :next call
72 :mess-up entry
86 ;;; This function handles merging the tail sets if CALL is potentially
87 ;;; tail-recursive, and is a call to a function with a different
88 ;;; TAIL-SET than CALL's FUN. This must be called whenever we alter
89 ;;; IR1 so as to place a local call in what might be a tail-recursive
90 ;;; context. Note that any call which returns its value to a RETURN is
91 ;;; considered potentially tail-recursive, since any implicit MV-PROG1
92 ;;; might be optimized away.
93 ;;;
94 ;;; We destructively modify the set for the calling function to
95 ;;; represent both, and then change all the functions in callee's set
96 ;;; to reference the first. If we do merge, we reoptimize the
97 ;;; RETURN-RESULT lvar to cause IR1-OPTIMIZE-RETURN to recompute the
98 ;;; tail set type.
112 t)))))
114 ;;; Convert a combination into a local call. We PROPAGATE-TO-ARGS, set
115 ;;; the combination kind to :LOCAL, add FUN to the CALLS of the
116 ;;; function that the call is in, call MERGE-TAIL-SETS, then replace
117 ;;; the function in the REF node with the new function.
118 ;;;
119 ;;; We change the REF last, since changing the reference can trigger
120 ;;; LET conversion of the new function, but will only do so if the
121 ;;; call is local. Note that the replacement may trigger LET
122 ;;; conversion or other changes in IR1. We must call MERGE-TAIL-SETS
123 ;;; with NEW-FUN before the substitution, since after the substitution
124 ;;; (and LET conversion), the call may no longer be recognizable as
125 ;;; tail-recursive.
139 \f
140 ;;;; external entry point creation
142 ;;; Return a LAMBDA form that can be used as the definition of the XEP
143 ;;; for FUN.
144 ;;;
145 ;;; If FUN is a LAMBDA, then we check the number of arguments
146 ;;; (conditional on policy) and call FUN with all the arguments.
147 ;;;
148 ;;; If FUN is an OPTIONAL-DISPATCH, then we dispatch off of the number
149 ;;; of supplied arguments by doing do an = test for each entry-point,
150 ;;; calling the entry with the appropriate prefix of the passed
151 ;;; arguments.
152 ;;;
153 ;;; If there is a &MORE arg, then there are a couple of optimizations
154 ;;; that we make (more for space than anything else):
155 ;;; -- If MIN-ARGS is 0, then we make the more entry a T clause, since
156 ;;; no argument count error is possible.
157 ;;; -- We can omit the = clause for the last entry-point, allowing the
158 ;;; case of 0 more args to fall through to the more entry.
159 ;;;
160 ;;; We don't bother to policy conditionalize wrong arg errors in
161 ;;; optional dispatches, since the additional overhead is negligible
162 ;;; compared to the cost of everything else going on.
163 ;;;
164 ;;; Note that if policy indicates it, argument type declarations in
165 ;;; FUN will be verified. Since nothing is known about the type of the
166 ;;; XEP arg vars, type checks will be emitted when the XEP's arg vars
167 ;;; are passed to the actual function.
190 ;; Force convertion of all entries
193 and n from min
197 ;; FIXME: Make sure that INDEX type distinguishes between
198 ;; target and host. (Probably just make the SB!XC:DEFTYPE
199 ;; different from CL:DEFTYPE.)
204 ;; KLUDGE: (NOT (< ...)) instead of >= avoids one round of
205 ;; deftransforms and lambda-conversion.
217 ;;; Make an external entry point (XEP) for FUN and return it. We
218 ;;; convert the result of MAKE-XEP-LAMBDA in the correct environment,
219 ;;; then associate this lambda with FUN as its XEP. After the
220 ;;; conversion, we iterate over the function's associated lambdas,
221 ;;; redoing local call analysis so that the XEP calls will get
222 ;;; converted.
223 ;;;
224 ;;; We set REANALYZE and REOPTIMIZE in the component, just in case we
225 ;;; discover an XEP after the initial local call analyze pass.
247 res)))
249 ;;; Notice a REF that is not in a local-call context. If the REF is
250 ;;; already to an XEP, then do nothing, otherwise change it to the
251 ;;; XEP, making an XEP if necessary.
252 ;;;
253 ;;; If REF is to a special :CLEANUP or :ESCAPE function, then we treat
254 ;;; it as though it was not an XEP reference (i.e. leave it alone).
262 \f
263 ;;; Attempt to convert all references to FUN to local calls. The
264 ;;; reference must be the function for a call, and the function lvar
265 ;;; must be used only once, since otherwise we cannot be sure what
266 ;;; function is to be called. The call lvar would be multiply used if
267 ;;; there is hairy stuff such as conditionals in the expression that
268 ;;; computes the function.
269 ;;;
270 ;;; If we cannot convert a reference, then we mark the referenced
271 ;;; function as an entry-point, creating a new XEP if necessary. We
272 ;;; don't try to convert calls that are in error (:ERROR kind.)
273 ;;;
274 ;;; This is broken off from LOCALL-ANALYZE-COMPONENT so that people
275 ;;; can force analysis of newly introduced calls. Note that we don't
276 ;;; do LET conversion here.
301 ;;; We examine all NEW-FUNCTIONALS in COMPONENT, attempting to convert
302 ;;; calls into local calls when it is legal. We also attempt to
303 ;;; convert each LAMBDA to a LET. LET conversion is also triggered by
304 ;;; deletion of a function reference, but functions that start out
305 ;;; eligible for conversion must be noticed sometime.
306 ;;;
307 ;;; Note that there is a lot of action going on behind the scenes
308 ;;; here, triggered by reference deletion. In particular, the
309 ;;; COMPONENT-LAMBDAS are being hacked to remove newly deleted and LET
310 ;;; converted LAMBDAs, so it is important that the LAMBDA is added to
311 ;;; the COMPONENT-LAMBDAS when it is. Also, the
312 ;;; COMPONENT-NEW-FUNCTIONALS may contain all sorts of drivel, since
313 ;;; it is not updated when we delete functions, etc. Only
314 ;;; COMPONENT-LAMBDAS is updated.
315 ;;;
316 ;;; COMPONENT-REANALYZE-FUNCTIONALS is treated similarly to
317 ;;; COMPONENT-NEW-FUNCTIONALS, but we don't add lambdas to the
318 ;;; LAMBDAS.
336 ;; Fix/check FUNCTIONAL's relationship to COMPONENT-LAMDBAS.
338 ;; Since FUNCTIONAL isn't a LAMBDA, this doesn't
339 ;; apply: no-op.
342 ; NEW-FUNCTIONALS, hence is new.
343 ;; FUNCTIONAL becomes part of COMPONENT-LAMBDAS now.
348 ;; FUNCTIONAL should be in COMPONENT-LAMBDAS already.
350 component)))))
361 ;; The original CMU CL code seemed to implicitly assume that
362 ;; COMPONENT is the only one here. Let's make that explicit.
373 ;;; If policy is auspicious and CALL is not in an XEP and we don't seem
374 ;;; to be in an infinite recursive loop, then change the reference to
375 ;;; reference a fresh copy. We return whichever function we decide to
376 ;;; reference.
394 original-functional)))))))
399 calls this LET-converted local function:~
400 ~% ~S"
404 expand because expansion refers to ~
405 the optimized away object ~S."
406 losing-local-object)))
413 original-functional)
416 converted-lambda))))
417 original-functional))
419 ;;; Dispatch to the appropriate function to attempt to convert a call.
420 ;;; REF must be a reference to a FUNCTIONAL. This is called in IR1
421 ;;; optimization as well as in local call analysis. If the call is is
422 ;;; already :LOCAL, we do nothing. If the call is already scheduled
423 ;;; for deletion, also do nothing (in addition to saving time, this
424 ;;; also avoids some problems with optimizing collections of functions
425 ;;; that are partially deleted.)
426 ;;;
427 ;;; This is called both before and after FIND-INITIAL-DFO runs. When
428 ;;; called on a :INITIAL component, we don't care whether the caller
429 ;;; and callee are in the same component. Afterward, we must stick
430 ;;; with whatever component division we have chosen.
431 ;;;
432 ;;; Before attempting to convert a call, we see whether the function
433 ;;; is supposed to be inline expanded. Call conversion proceeds as
434 ;;; before after any expansion.
435 ;;;
436 ;;; We bind *COMPILER-ERROR-CONTEXT* to the node for the call so that
437 ;;; warnings will get the right context.
452 component))))
455 original-fun))
473 ;;; Attempt to convert a multiple-value call. The only interesting
474 ;;; case is a call to a function that LOOKS-LIKE-AN-MV-BIND, has
475 ;;; exactly one reference and no XEP, and is called with one values
476 ;;; lvar.
477 ;;;
478 ;;; We change the call to be to the last optional entry point and
479 ;;; change the call to be local. Due to our preconditions, the call
480 ;;; should eventually be converted to a let, but we can't do that now,
481 ;;; since there may be stray references to the e-p lambda due to
482 ;;; optional defaulting code.
483 ;;;
484 ;;; We also use variable types for the called function to construct an
485 ;;; assertion for the values lvar.
486 ;;;
487 ;;; See CONVERT-CALL for additional notes on MERGE-TAIL-SETS, etc.
510 ;;; Attempt to convert a call to a lambda. If the number of args is
511 ;;; wrong, we give a warning and mark the call as :ERROR to remove it
512 ;;; from future consideration. If the argcount is O.K. then we just
513 ;;; convert it.
522 'local-argument-mismatch
523 :format-control
524 "function called with ~R argument~:P, but wants exactly ~R"
527 \f
528 ;;;; &OPTIONAL, &MORE and &KEYWORD calls
530 ;;; This is similar to CONVERT-LAMBDA-CALL, but deals with
531 ;;; OPTIONAL-DISPATCHes. If only fixed args are supplied, then convert
532 ;;; a call to the correct entry point. If &KEY args are supplied, then
533 ;;; dispatch to a subfunction. We don't convert calls to functions
534 ;;; that have a &MORE (or &REST) arg.
543 'local-argument-mismatch
544 :format-control
545 "function called with ~R argument~:P, but wants at least ~R"
557 'local-argument-mismatch
558 :format-control
559 "function called with ~R argument~:P, but wants at most ~R"
560 :format-arguments
565 ;;; This function is used to convert a call to an entry point when
566 ;;; complex transformations need to be done on the original arguments.
567 ;;; ENTRY is the entry point function that we are calling. VARS is a
568 ;;; list of variable names which are bound to the original call
569 ;;; arguments. IGNORES is the subset of VARS which are ignored. ARGS
570 ;;; is the list of arguments to the entry point function.
571 ;;;
572 ;;; In order to avoid gruesome graph grovelling, we introduce a new
573 ;;; function that rearranges the arguments and calls the entry point.
574 ;;; We analyze the new function and the entry point immediately so
575 ;;; that everything gets converted during the single pass.
592 ;;; Use CONVERT-HAIRY-FUN-ENTRY to convert a &MORE-arg call to a known
593 ;;; function into a local call to the MAIN-ENTRY.
594 ;;;
595 ;;; First we verify that all keywords are constant and legal. If there
596 ;;; aren't, then we warn the user and don't attempt to convert the call.
597 ;;;
598 ;;; We massage the supplied &KEY arguments into the order expected
599 ;;; by the main entry. This is done by binding all the arguments to
600 ;;; the keyword call to variables in the introduced lambda, then
601 ;;; passing these values variables in the correct order when calling
602 ;;; the main entry. Unused arguments (such as the keywords themselves)
603 ;;; are discarded simply by not passing them along.
604 ;;;
605 ;;; If there is a &REST arg, then we bundle up the args and pass them
606 ;;; to LIST.
628 (:keyword
691 (:optional
695 (:rest
698 (:keyword
716 \f
717 ;;;; LET conversion
718 ;;;;
719 ;;;; Converting to a LET has differing significance to various parts
720 ;;;; of the compiler:
721 ;;;; -- The body of a LET is spliced in immediately after the
722 ;;;; corresponding combination node, making the control transfer
723 ;;;; explicit and allowing LETs to be mashed together into a single
724 ;;;; block. The value of the LET is delivered directly to the
725 ;;;; original lvar for the call, eliminating the need to
726 ;;;; propagate information from the dummy result lvar.
727 ;;;; -- As far as IR1 optimization is concerned, it is interesting in
728 ;;;; that there is only one expression that the variable can be bound
729 ;;;; to, and this is easily substituted for.
730 ;;;; -- LETs are interesting to environment analysis and to the back
731 ;;;; end because in most ways a LET can be considered to be "the
732 ;;;; same function" as its home function.
733 ;;;; -- LET conversion has dynamic scope implications, since control
734 ;;;; transfers within the same environment are local. In a local
735 ;;;; control transfer, cleanup code must be emitted to remove
736 ;;;; dynamic bindings that are no longer in effect.
738 ;;; Set up the control transfer to the called CLAMBDA. We split the
739 ;;; call block immediately after the call, and link the head of
740 ;;; CLAMBDA to the call block. The successor block after splitting
741 ;;; (where we return to) is returned.
742 ;;;
743 ;;; If the lambda is is a different component than the call, then we
744 ;;; call JOIN-COMPONENTS. This only happens in block compilation
745 ;;; before FIND-INITIAL-DFO.
762 next-block)))
764 ;;; Remove CLAMBDA from the tail set of anything it used to be in the
765 ;;; same set as; but leave CLAMBDA with a valid tail set value of
766 ;;; its own, for the benefit of code which might try to pull
767 ;;; something out of it (e.g. return type).
769 ;; Until sbcl-0.pre7.37.flaky5.2, we did
770 ;; (LET ((TAILS (LAMBDA-TAIL-SET CLAMBDA)))
771 ;; (SETF (TAIL-SET-FUNS TAILS)
772 ;; (DELETE CLAMBDA (TAIL-SET-FUNS TAILS))))
773 ;; (SETF (LAMBDA-TAIL-SET CLAMBDA) NIL)
774 ;; here. Apparently the idea behind the (SETF .. NIL) was that since
775 ;; TAIL-SET-FUNS no longer thinks we're in the tail set, it's
776 ;; inconsistent, and perhaps unsafe, for us to think we're in the
777 ;; tail set. Unfortunately..
778 ;;
779 ;; The (SETF .. NIL) caused problems in sbcl-0.pre7.37.flaky5.2 when
780 ;; I was trying to get Python to emit :EXTERNAL LAMBDAs directly
781 ;; (instead of only being able to emit funny little :TOPLEVEL stubs
782 ;; which you called in order to get the address of an external LAMBDA):
783 ;; the external function was defined in terms of internal function,
784 ;; which was LET-converted, and then things blew up downstream when
785 ;; FINALIZE-XEP-DEFINITION tried to find out its DEFINED-TYPE from
786 ;; the now-NILed-out TAIL-SET. So..
787 ;;
788 ;; To deal with this problem, we no longer NIL out
789 ;; (LAMBDA-TAIL-SET CLAMBDA) here. Instead:
790 ;; * If we're the only function in TAIL-SET-FUNS, it should
791 ;; be safe to leave ourself linked to it, and it to you.
792 ;; * If there are other functions in TAIL-SET-FUNS, then we're
793 ;; afraid of future optimizations on those functions causing
794 ;; the TAIL-SET object no longer to be valid to describe our
795 ;; return value. Thus, we delete ourselves from that object;
796 ;; but we save a newly-allocated tail-set, derived from the old
797 ;; one, for ourselves, for the use of later code (e.g.
798 ;; FINALIZE-XEP-DEFINITION) which might want to
799 ;; know about our return type.
808 ;; The documentation on TAIL-SET-INFO doesn't tell whether it could
809 ;; remain valid in this case, so we nuke it on the theory that
810 ;; missing information tends to be less dangerous than incorrect
811 ;; information.
814 ;;; Handle the PHYSENV semantics of LET conversion. We add CLAMBDA and
815 ;;; its LETs to LETs for the CALL's home function. We merge the calls
816 ;;; for CLAMBDA with the calls for the home function, removing CLAMBDA
817 ;;; in the process. We also merge the ENTRIES.
818 ;;;
819 ;;; We also unlink the function head from the component head and set
820 ;;; COMPONENT-REANALYZE to true to indicate that the DFO should be
821 ;;; recomputed.
841 ;; CLAMBDA belongs to HOME now.
851 ;; All of CLAMBDA's LETs belong to HOME now.
857 ;; CLAMBDA no longer has an independent existence as an entity
858 ;; which has LETs.
861 ;; HOME no longer calls CLAMBDA, and owns all of CLAMBDA's old
862 ;; DFO dependencies.
866 ;; CLAMBDA no longer has an independent existence as an entity
867 ;; which calls things or has DFO dependencies.
870 ;; All of CLAMBDA's ENTRIES belong to HOME now.
874 ;; CLAMBDA no longer has an independent existence as an entity
875 ;; with ENTRIES.
880 ;;; Handle the value semantics of LET conversion. Delete FUN's return
881 ;;; node, and change the control flow to transfer to NEXT-BLOCK
882 ;;; instead. Move all the uses of the result lvar to CALL's lvar.
901 ;; FIXME: Replace the call with unsafe CAST. -- APD, 2003-01-26
908 ;;; We are converting FUN to be a LET when the call is in a non-tail
909 ;;; position. Any previously tail calls in FUN are no longer tail
910 ;;; calls, and must be restored to normal calls which transfer to
911 ;;; NEXT-BLOCK (FUN's return point.) We can't do this by DO-USES on
912 ;;; the RETURN-RESULT, because the return might have been deleted (if
913 ;;; all calls were TR.)
932 ;; The called function might be an assignment in the
933 ;; case where we are currently converting that function.
934 ;; In steady-state, assignments never appear as a called
935 ;; function.
936 (:assignment
940 ;;; Deal with returning from a LET or assignment that we are
941 ;;; converting. FUN is the function we are calling, CALL is a call to
942 ;;; FUN, and NEXT-BLOCK is the return point for a non-tail call, or
943 ;;; NULL if call is a tail call.
944 ;;;
945 ;;; If the call is not a tail call, then we must do
946 ;;; UNCONVERT-TAIL-CALLS, since a tail call is a call which returns
947 ;;; its value out of the enclosing non-let function. When call is
948 ;;; non-TR, we must convert it back to an ordinary local call, since
949 ;;; the value must be delivered to the receiver of CALL's value.
950 ;;;
951 ;;; We do different things depending on whether the caller and callee
952 ;;; have returns left:
954 ;;; -- If the callee has no return we just do MOVE-LET-CALL-CONT.
955 ;;; Either the function doesn't return, or all returns are via
956 ;;; tail-recursive local calls.
957 ;;; -- If CALL is a non-tail call, or if both have returns, then
958 ;;; we delete the callee's return, move its uses to the call's
959 ;;; result lvar, and transfer control to the appropriate
960 ;;; return point.
961 ;;; -- If the callee has a return, but the caller doesn't, then we
962 ;;; move the return to the caller.
991 ;;; Actually do LET conversion. We call subfunctions to do most of the
992 ;;; work. We do REOPTIMIZE-LVAR on the args and CALL's lvar so that
993 ;;; LET-specific IR1 optimizations get a chance. We blow away any
994 ;;; entry for the function in *FREE-FUNS* so that nobody will create
995 ;;; new references to it.
1000 nil
1001 next-block)))
1005 ;; If CALL has a derive type NIL, it means that "its return" is
1006 ;; unreachable, but the next BIND is still reachable; in order to
1007 ;; not confuse MAYBE-TERMINATE-BLOCK...
1010 ;;; Reoptimize all of CALL's args and its result.
1019 ;;; Are there any declarations in force to say CLAMBDA shouldn't be
1020 ;;; LET converted?
1022 ;; From the user's point of view, LET-converting something that
1023 ;; has a name is inlining it. (The user can't see what we're doing
1024 ;; with anonymous things, and suppressing inlining
1025 ;; for such things can easily give Python acute indigestion, so
1026 ;; we don't.)
1028 ;; ANSI requires that explicit NOTINLINE be respected.
1030 ;; If (= LET-CONVERSION 0) we can guess that inlining
1031 ;; generally won't be appreciated, but if the user
1032 ;; specifically requests inlining, that takes precedence over
1033 ;; our general guess.
1037 ;;; We also don't convert calls to named functions which appear in the
1038 ;;; initial component, delaying this until optimization. This
1039 ;;; minimizes the likelihood that we will LET-convert a function which
1040 ;;; may have references added due to later local inline expansion.
1047 ;;; This function is called when there is some reason to believe that
1048 ;;; CLAMBDA might be converted into a LET. This is done after local
1049 ;;; call analysis, and also when a reference is deleted. We return
1050 ;;; true if we converted.
1055 ;; We only convert to a LET when the function is a normal local
1056 ;; function, has no XEP, and is referenced in exactly one local
1057 ;; call. Conversion is also inhibited if the only reference is in
1058 ;; a block about to be deleted.
1059 ;;
1060 ;; These rules limiting LET conversion may seem unnecessarily
1061 ;; restrictive, since there are some cases where we could do the
1062 ;; return with a jump that don't satisfy these requirements. The
1063 ;; reason for doing things this way is that it makes the concept
1064 ;; of a LET much more useful at the level of IR1 semantics. The
1065 ;; :ASSIGNMENT function kind provides another way to optimize
1066 ;; calls to single-return/multiple call functions.
1067 ;;
1068 ;; We don't attempt to convert calls to functions that have an
1069 ;; XEP, since we might be embarrassed later when we want to
1070 ;; convert a newly discovered local call. Also, see
1071 ;; OK-INITIAL-CONVERT-P.
1088 nil)))
1097 t))))
1098 \f
1099 ;;;; tail local calls and assignments
1101 ;;; Return T if there are no cleanups between BLOCK1 and BLOCK2, or if
1102 ;;; they definitely won't generate any cleanup code. Currently we
1103 ;;; recognize lexical entry points that are only used locally (if at
1104 ;;; all).
1118 ;;; If a potentially TR local call really is TR, then convert it to
1119 ;;; jump directly to the called function. We also call
1120 ;;; MAYBE-CONVERT-TO-ASSIGNMENT. The first value is true if we
1121 ;;; tail-convert. The second is the value of M-C-T-A.
1128 ;; this is a tail call
1132 ;; If the call is in an XEP, we might decide to make it
1133 ;; non-tail so that we can use known return inside the
1134 ;; component.
1146 ;;; This is called when we believe it might make sense to convert
1147 ;;; CLAMBDA to an assignment. All this function really does is
1148 ;;; determine when a function with more than one call can still be
1149 ;;; combined with the calling function's environment. We can convert
1150 ;;; when:
1151 ;;; -- The function is a normal, non-entry function, and
1152 ;;; -- Except for one call, all calls must be tail recursive calls
1153 ;;; in the called function (i.e. are self-recursive tail calls)
1154 ;;; -- OK-INITIAL-CONVERT-P is true.
1155 ;;;
1156 ;;; There may be one outside call, and it need not be tail-recursive.
1157 ;;; Since all tail local calls have already been converted to direct
1158 ;;; transfers, the only control semantics needed are to splice in the
1159 ;;; body at the non-tail call. If there is no non-tail call, then we
1160 ;;; need only merge the environments. Both cases are handled by
1161 ;;; LET-CONVERT.
1162 ;;;
1163 ;;; ### It would actually be possible to allow any number of outside
1164 ;;; calls as long as they all return to the same place (i.e. have the
1165 ;;; same conceptual continuation.) A special case of this would be
1166 ;;; when all of the outside calls are tail recursive.
1193 t)
1195 nil))))))