bug#37758: 27.0.50; Support multibyte characters in function 'auth-source--pad'an; with patch

bug#37758: 27.0.50; Support multibyte characters in function 'auth-source--pad'an; with patch

[mgcyung]

[-- Attachment #1: Type: text/plain, Size: 425 bytes --]


'auth-source--pad' use 'length' to get byte length of a
string. However, 'length' return the number of characters which is not
the byte length with multibyte characters such as Chinese
character. This may causes an error "GnuTLS cipher
AES-256-CBC/encrypt input block length xxxx is not a multiple of the
required 16".

According to the doc of 'length'. 'length' can be replaced by
'string-bytes'. And a patch is attached.


[-- Attachment #2: auth-source--pad_byte-length.patch --]
[-- Type: application/octet-stream, Size: 817 bytes --]

diff --git a/lisp/auth-source.el b/lisp/auth-source.el
index 9061d41556..80559b8f47 100644
--- a/lisp/auth-source.el
+++ b/lisp/auth-source.el
@@ -1190,13 +1190,13 @@ FILE is the file from which we obtained this token."
 
 (defun auth-source--pad (string length)
   "Pad string S to a modulo of LENGTH."
-  (let ((pad (- length (mod (length string) length))))
+  (let ((pad (- length (mod (string-bytes string) length))))
     (concat string (make-string pad pad))))
 
 (defun auth-source--unpad (string)
   "Remove PKCS#7 padding from STRING."
-  (substring string 0 (- (length string)
-			 (aref string (1- (length string))))))
+  (substring string 0 (- (string-bytes string)
+			 (aref string (1- (string-bytes string))))))
 
 (defun auth-source--deobfuscate (data)
   (if (and (fboundp 'gnutls-symmetric-encrypt)

[-- Attachment #3: Type: text/plain, Size: 2 bytes --]

bug#37758: 27.0.50; Support multibyte characters in function 'auth-source--pad'an; with patch

[Lars Ingebrigtsen]

mgcyung <mgcyung@gmail.com> writes:

> 'auth-source--pad' use 'length' to get byte length of a
> string. However, 'length' return the number of characters which is not
> the byte length with multibyte characters such as Chinese
> character. This may causes an error "GnuTLS cipher
> AES-256-CBC/encrypt input block length xxxx is not a multiple of the
> required 16".
>
> According to the doc of 'length'. 'length' can be replaced by
> 'string-bytes'. And a patch is attached.

Thanks for the patch, but I don't think it's a quite safe change.
Instead the string should be made into a sequence of bytes before
padding (and converted back again afterwards) to ensure that it survives
the encryption, I think.

I've now made this change on the trunk.  Could you check that it works
for you?

-- 
(domestic pets only, the antidote for overdose, milk.)
   bloggy blog: http://lars.ingebrigtsen.no

bug#37758: 27.0.50; Support multibyte characters in function 'auth-source--pad'an; with patch

[mgcyung]

Yes. It works fine.

On Tue, 15 Oct 2019 14:43:58 +0800,
Lars Ingebrigtsen wrote:
> 
> mgcyung <mgcyung@gmail.com> writes:
> 
> > 'auth-source--pad' use 'length' to get byte length of a
> > string. However, 'length' return the number of characters which is not
> > the byte length with multibyte characters such as Chinese
> > character. This may causes an error "GnuTLS cipher
> > AES-256-CBC/encrypt input block length xxxx is not a multiple of the
> > required 16".
> >
> > According to the doc of 'length'. 'length' can be replaced by
> > 'string-bytes'. And a patch is attached.
> 
> Thanks for the patch, but I don't think it's a quite safe change.
> Instead the string should be made into a sequence of bytes before
> padding (and converted back again afterwards) to ensure that it survives
> the encryption, I think.
> 
> I've now made this change on the trunk.  Could you check that it works
> for you?
> 
> -- 
> (domestic pets only, the antidote for overdose, milk.)
>    bloggy blog: http://lars.ingebrigtsen.no

bug#37758: 27.0.50; Support multibyte characters in function 'auth-source--pad'an; with patch

[Lars Ingebrigtsen]

mgcyung <mgcyung@gmail.com> writes:

> Yes. It works fine.

Thanks for checking; closing this bug report.

-- 
(domestic pets only, the antidote for overdose, milk.)
   bloggy blog: http://lars.ingebrigtsen.no