X-Git-Url: http://repo.macrolet.net/gitweb/?a=blobdiff_plain;f=src%2Fcode%2Fhost-c-call.lisp;h=1976a2f847da2eaa64f0c873d02c4b7748ee52b5;hb=1600081cf1b71b3d0e2e40de1c1c124a3a4fd40c;hp=409fd9fa8013890c280afa886cc0695429a15d13;hpb=988afd9d54ba6c8a915544822658824ab6ae0d6c;p=sbcl.git diff --git a/src/code/host-c-call.lisp b/src/code/host-c-call.lisp index 409fd9f..1976a2f 100644 --- a/src/code/host-c-call.lisp +++ b/src/code/host-c-call.lisp @@ -11,62 +11,111 @@ (/show0 "host-c-call.lisp 12") -(define-alien-type-class (c-string :include pointer :include-args (to))) +(define-alien-type-class (c-string :include pointer :include-args (to)) + (external-format :default :type keyword) + (element-type 'character :type (member character base-char))) -(define-alien-type-translator c-string () +(define-alien-type-translator c-string + (&key (external-format :default) + (element-type 'character)) (make-alien-c-string-type - :to (parse-alien-type 'char (sb!kernel:make-null-lexenv)))) + :to (parse-alien-type 'char (sb!kernel:make-null-lexenv)) + :element-type element-type + :external-format external-format)) + +(defun c-string-external-format (type) + (let ((external-format (alien-c-string-type-external-format type))) + (if (eq external-format :default) + (default-c-string-external-format) + external-format))) (define-alien-type-method (c-string :unparse) (type) - (declare (ignore type)) - 'c-string) + (list 'c-string + :external-format (alien-c-string-type-external-format type) + :element-type (alien-c-string-type-element-type type))) (define-alien-type-method (c-string :lisp-rep) (type) (declare (ignore type)) '(or simple-string null (alien (* char)))) +(defun c-string-needs-conversion-p (type) + #+sb-xc-host + t + #-sb-xc-host + (let ((external-format (sb!impl::get-external-format + ;; Can't use C-STRING-EXTERNAL-FORMAT here, + ;; since the meaning of :DEFAULT can change + ;; when *DEFAULT-C-STRING-EXTERNAL-FORMAT* + ;; changes. + (alien-c-string-type-external-format type)))) + (not (and external-format + (or (eq (caar external-format) :ascii) + ;; On non-SB-UNICODE all latin-1 codepoints will fit + ;; into a base-char, on SB-UNICODE they won't. + #!-sb-unicode + (eq (caar external-format) :latin-1)))))) + (define-alien-type-method (c-string :naturalize-gen) (type alien) - (declare (ignore type)) `(if (zerop (sap-int ,alien)) nil - (%naturalize-c-string ,alien))) + ;; Check whether we need to do a full external-format + ;; conversion, or whether we can just do a cheap byte-by-byte + ;; copy of the c-string data. + ;; + ;; On SB-UNICODE we can never do the cheap copy, even if the + ;; external format and element-type are suitable, since + ;; simple-base-strings may not contain ISO-8859-1 characters. + ;; If we need to check for non-ascii data in the input, we + ;; might as well go through the usual external-format machinery + ;; instead of rewriting another version of it. + ,(if #!+sb-unicode t + #!-sb-unicode (c-string-needs-conversion-p type) + `(sb!alien::c-string-to-string ,alien + (c-string-external-format ,type) + (alien-c-string-type-element-type + ,type)) + `(%naturalize-c-string ,alien)))) (define-alien-type-method (c-string :deport-gen) (type value) - (declare (ignore type)) - `(etypecase ,value - (null (int-sap 0)) - ((alien (* char)) (alien-sap ,value)) - (simple-base-string (vector-sap ,value)) - (simple-string (vector-sap (coerce ,value 'simple-base-string))))) - -(/show0 "host-c-call.lisp 42") - -(define-alien-type-class (utf8-string :include pointer :include-args (to))) - -(define-alien-type-translator utf8-string () - (make-alien-utf8-string-type - :to (parse-alien-type 'char (sb!kernel:make-null-lexenv)))) - -(define-alien-type-method (utf8-string :unparse) (type) - (declare (ignore type)) - 'utf8-string) - -(define-alien-type-method (utf8-string :lisp-rep) (type) - (declare (ignore type)) - '(or simple-string null (alien (* char)))) - -(define-alien-type-method (utf8-string :naturalize-gen) (type alien) - (declare (ignore type)) - `(if (zerop (sap-int ,alien)) - nil - (%naturalize-utf8-string ,alien))) - -(define-alien-type-method (utf8-string :deport-gen) (type value) - (declare (ignore type)) `(etypecase ,value (null (int-sap 0)) ((alien (* char)) (alien-sap ,value)) - (simple-base-string (vector-sap ,value)) - (simple-string (vector-sap (%deport-utf8-string ,value))))) + ;; FIXME: GC safety alert! These SAPs are not safe, since the + ;; Lisp string can move. This is not hard to arrange, for example + ;; the following will fail very quickly on a SB-UNICODE build: + ;; + ;; (setf (bytes-consed-between-gcs) 4096) + ;; (define-alien-routine "strcmp" int (s1 c-string) (s2 c-string)) + ;; + ;; (loop + ;; (let ((string "hello, world")) + ;; (assert (zerop (strcmp string string))))) + ;; + ;; (This will appear to work on post-0.9.8.19 GENCGC, since + ;; the GC no longer zeroes memory immediately after releasing + ;; it after a minor GC. Either enabling the READ_PROTECT_FREE_PAGES + ;; #define in gencgc.c or modifying the example so that a major + ;; GC will occasionally be triggered would unmask the bug). + ;; + ;; The pure VECTOR-SAP branch for the SIMPLE-BASE-STRING case + ;; will generally be very hard to trigger on GENCGC (even when + ;; threaded) thanks to GC conservativeness. It's mostly a problem + ;; on cheneygc. -- JES, 2006-01-13 + (simple-base-string + ,(if (c-string-needs-conversion-p type) + ;; If the alien type is not ascii-compatible (+SB-UNICODE) + ;; or latin-1-compatible (-SB-UNICODE), we need to do + ;; external format conversion. + `(vector-sap (string-to-c-string ,value + (c-string-external-format ,type))) + ;; Otherwise we can just pass it uncopied. + `(vector-sap ,value))) + ;; This case, on the other hand, will cause trouble on GENCGC, since + ;; we're taking the SAP of a immediately discarded temporary -> the + ;; conservativeness doesn't protect us. + ;; -- JES, 2006-01-13 + (simple-string + (vector-sap (string-to-c-string ,value + (c-string-external-format ,type)))))) (/show0 "host-c-call.lisp end of file")