On Sun, 21 May 2000, Petr Vandrovec wrote:
> Hi Urban,
> from ncpfs point of view it looks OK. I have one question - currently
> callers of ncp_io2vol conversions assume that this conversion will not
> produce string longer than original (it can produce shorter if external
> encoding is utf8). But after multibyte encodings come to kernel, this
> may be no longer true. Is there any method to compute upper boundary of
> resulting string size? I do not want to use get_free_page() for
> temporary results :-(
You would like something
nls->max_character_size
or
nls->strlen(unicode_str)
or
nls->str_ubound(str_len)
or
#define NLS_MAX_CODEPAGE_CHAR_SIZE 2
?
I don't think there is anything in the current interface to ask the nls
how long strings it may produce.
You allocate these temporary results on the stack ... how long can 'len'
be?
- __u8 __name[len];
+ __u8 __name[2*len]; /* 16-bit character support hack :) */
For my smbfs-nls thing I now have a temporary buffer in the 'server'
structure. That works because the server struct is locked and there is no
recursion. But it doesn't adjust the maxlength by which encoding we use
and will give errors if the name is too long (hmm, I think it will
truncate them silently ... oops :)
Also, iirc the linux side encoding, as returned to userspace, is supposed
to be (capable of handling) utf8, not just iso-whatever. Currently that
means that 6-byte characters are possible (nls_base.c).
The ncpfs code I changed will complain (loudly?) if the charset wants more
than one byte per character.
if (out->uni2char(uc.uni2, uc.uni1, &nc, 1, &len) == -1)
return -EINVAL;
The 1 is for the maximum number of characters it is allowed to write, and
the 2.2.16-pre code has a similar limitation. Since you do toupper/tolower
that needs fixing too for the new codepages.
/Urban
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/
This archive was generated by hypermail 2b29 : Sun May 21 2000 - 07:39:16 EDT