Mercurial > hg > xemacs-beta

diff tests/automated/lisp-tests.el @ 5892:053ef01b71a8

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Import the #'clear-string API from GNU, use it in tls.c src/ChangeLog addition: 2015-04-18 Aidan Kehoe <kehoea@parhasard.net> * sequence.c (Fclear_string): New, API from GNU. Zero a string's contents, making sure the text is not kept around even when the string's data is reallocated because of a changed character length. * sequence.c (syms_of_sequence): Make it available to Lisp. * lisp.h: Make it available to C code. * tls.c (nss_pk11_password): Use it. * tls.c (gnutls_pk11_password): Use it. * tls.c (openssl_password): Use it. tests/ChangeLog addition: 2015-04-18 Aidan Kehoe <kehoea@parhasard.net> * automated/lisp-tests.el: Test #'clear-string, just added. Unfortunately there's no way to be certain from Lisp that the old password data has been erased after realloc; it may be worth adding a test to tests.c, but *we'll be reading memory we shouldn't be*, so that gives me pause.
author Aidan Kehoe <kehoea@parhasard.net>
date Sat, 18 Apr 2015 23:00:14 +0100
parents bd644055ef44
children d3d073aceaea
line wrap: on
line diff
--- a/tests/automated/lisp-tests.el	Sat Apr 11 18:34:14 2015 +0100
+++ b/tests/automated/lisp-tests.el	Sat Apr 18 23:00:14 2015 +0100
@@ -3796,6 +3796,35 @@
   (Assert (eql ?\x09 (digit-char 9 nil binary-table))
           "checking `digit-char' reflects RADIX-TABLE, 9, base 10"))
 
+;; Test #'clear-string.
+
+(Check-Error wrong-type-argument (clear-string [?\x00 ?\xff]))
+(Check-Error wrong-type-argument (clear-string '(?\x00 ?\xff)))
+(Check-Error wrong-type-argument (clear-string #*1010))
+(Check-Error wrong-number-of-arguments (clear-string "hello" ?*))
+
+(let* ((template (concat
+                  "this is a template string, "
+                  (if (unicode-to-char #x06af)
+                      (decode-coding-string
+                       (concat
+                        "\xd8\xa8\xd9\x87 \xd9\x86\xd8\xb8\xd8\xb1 "
+                        "\xd9\x85\xd9\x86 \xd8\xae\xd9\x88\xda\xa9 "
+                        "\xd8\xae\xd9\x88\xd8\xb4\xd9\x85\xd8\xb2\xd9\x87 "
+                        "\xd8\xa7\xd8\xb3\xd8\xaa")
+                       'utf-8))))
+       (length (length template))
+       (null (make-string length ?\x00)))
+  (Assert (null (clear-string (copy-sequence template))))
+  (Assert (eql length (let ((string (copy-sequence template)))
+                        (clear-string string)
+                        (length string))))
+  (Assert (equal null (let ((string (copy-sequence template)))
+                        (clear-string string)
+                        string))))
+
+;; No way to check from Lisp whether the data was actually nulled.
+
 ;; Check that a bug in #'check-type with non-setfable PLACE (something not
 ;; actually specified by Common Lisp) has been fixed.
 (Assert (prog1 t (check-type 300 fixnum))