(do-blocks (block component)
(cond
((or (block-delete-p block)
- (null (block-pred block))
- (eq (functional-kind (block-home-lambda block)) :deleted))
+ (null (block-pred block)))
(delete-block block))
+ ((eq (functional-kind (block-home-lambda block)) :deleted)
+ ;; Preserve the BLOCK-SUCC invariant that almost every block has
+ ;; one successor (and a block with DELETE-P set is an acceptable
+ ;; exception).
+ (labels ((mark-blocks (block)
+ (dolist (pred (block-pred block))
+ (when (and (not (block-delete-p pred))
+ (eq (functional-kind (block-home-lambda pred))
+ :deleted))
+ (setf (block-delete-p pred) t)
+ (mark-blocks pred)))))
+ (mark-blocks block)
+ (delete-block block)))
(t
(loop
(let ((succ (block-succ block)))
(unless (and succ (null (rest succ)))
(return)))
-
+
(let ((last (block-last block)))
(typecase last
(cif
(exit
(when (maybe-delete-exit last)
(return)))))
-
- (unless (join-successor-if-possible block)
+
+ (unless (join-successor-if-possible block)
(return)))
(when (and (block-reoptimize block) (block-component block))
(let ((args (mapcar #'continuation-value (combination-args call)))
(fun-name (combination-fun-source-name call)))
(multiple-value-bind (values win)
- (careful-call fun-name args call "constant folding")
+ (careful-call fun-name
+ args
+ call
+ ;; Note: CMU CL had COMPILER-WARN here, and that
+ ;; seems more natural, but it's probably not.
+ ;;
+ ;; It's especially not while bug 173 exists:
+ ;; Expressions like
+ ;; (COND (END
+ ;; (UNLESS (OR UNSAFE? (<= END SIZE)))
+ ;; ...))
+ ;; can cause constant-folding TYPE-ERRORs (in
+ ;; #'<=) when END can be proved to be NIL, even
+ ;; though the code is perfectly legal and safe
+ ;; because a NIL value of END means that the
+ ;; #'<= will never be executed.
+ ;;
+ ;; Moreover, even without bug 173,
+ ;; quite-possibly-valid code like
+ ;; (COND ((NONINLINED-PREDICATE END)
+ ;; (UNLESS (<= END SIZE))
+ ;; ...))
+ ;; (where NONINLINED-PREDICATE is something the
+ ;; compiler can't do at compile time, but which
+ ;; turns out to make the #'<= expression
+ ;; unreachable when END=NIL) could cause errors
+ ;; when the compiler tries to constant-fold (<=
+ ;; END SIZE).
+ ;;
+ ;; So, with or without bug 173, it'd be
+ ;; unnecessarily evil to do a full
+ ;; COMPILER-WARNING (and thus return FAILURE-P=T
+ ;; from COMPILE-FILE) for legal code, so we we
+ ;; use a wimpier COMPILE-STYLE-WARNING instead.
+ #'compiler-style-warn
+ "constant folding")
(if (not win)
(setf (combination-kind call) :error)
(let ((dummies (make-gensym-list (length args))))
(derive-node-type node (continuation-type (set-value node)))
(values))
-;;; Return true if the value of Ref will always be the same (and is
+;;; Return true if the value of REF will always be the same (and is
;;; thus legal to substitute.)
(defun constant-reference-p (ref)
(declare (type ref ref))