[PATCH v4 02/11] lib/charset: add u16_strlcat() function

[Masahisa Kojima]

On Sat, 16 Apr 2022 at 16:32, Heinrich Schuchardt <xypron.glpk at gmx.de> wrote:
>
> On 3/24/22 14:54, Masahisa Kojima wrote:
> > Provide u16 string version of strlcat().
> >
> > Signed-off-by: Masahisa Kojima <masahisa.kojima at linaro.org>
> > Reviewed-by: Simon Glass <sjg at chromium.org>
> > ---
> > Changes in v4:
> > - add blank line above the return statement
> >
> > Changes in v2:
> > - implement u16_strlcat(with the destination buffer size in argument)
> >    instead of u16_strcat
> >
> >   include/charset.h | 15 +++++++++++++++
> >   lib/charset.c     | 21 +++++++++++++++++++++
> >   2 files changed, 36 insertions(+)
> >
> > diff --git a/include/charset.h b/include/charset.h
> > index b93d023092..dc5fc275ec 100644
> > --- a/include/charset.h
> > +++ b/include/charset.h
> > @@ -259,6 +259,21 @@ u16 *u16_strcpy(u16 *dest, const u16 *src);
> >    */
> >   u16 *u16_strdup(const void *src);
> >
> > +/**
> > + * u16_strlcat() - Append a length-limited, %NUL-terminated string to another
> > + *
> > + * Append the src string to the dest string, overwriting the terminating
> > + * null word at the end of dest, and then adds a terminating null word.
> > + * It will append at most size - u16_strlen(dst) - 1 bytes, NUL-terminating the result.
>
> Why "- 1"?

It is my mistake, it should be 2.

>
> If size is even, we append up to size - u16_strlen(dst) - 2 bytes. The
> two extra bytes used for 0x0000.
> If size is odd, we append up to size - u16_strlen(dst) - 3 bytes leaving
> one byte of the buffer unused.

Thanks, It clearly explains the behavior.

>
> > + *
> > + * @dest:            destination buffer (null terminated)
> > + * @src:             source buffer (null terminated)
> > + * @size:            destination buffer size in bytes
>
> s/$/ including the trailing 0x0000/

OK, I will update "(null terminated)" to the suggested one.

>
> > + * Return:           total size of the created string in bytes.
> > + *                   If return value >= size, truncation occurred.
> > + */
> > +size_t u16_strlcat(u16 *dest, const u16 *src, size_t size);
> > +
> >   /**
> >    * utf16_to_utf8() - Convert an utf16 string to utf8
> >    *
> > diff --git a/lib/charset.c b/lib/charset.c
> > index f44c58d9d8..47997eca7d 100644
> > --- a/lib/charset.c
> > +++ b/lib/charset.c
> > @@ -428,6 +428,27 @@ u16 *u16_strdup(const void *src)
> >       return new;
> >   }
> >
> > +size_t u16_strlcat(u16 *dest, const u16 *src, size_t size)
> > +{
>
> If you start the function with
>
>      size >>= 1;
>
> or
>
>      size /= sizeof(u16);
>
> this might simplify the code.

In u16_strlcat(), there are two size definitions, u16 string size and
buffer size.
I will rename some of the variables to clearly identify the meaning.

>
> > +     size_t dstrlen = u16_strnlen(dest, size >> 1);
> > +     size_t dlen = dstrlen * sizeof(u16);
> > +     size_t len = u16_strlen(src) * sizeof(u16);
> > +     size_t ret = dlen + len;
>
> This misses the  trailing 0x0000.

Strlcat() is not the C standard function, but the linux implementation
of strlcat() does not include trailing 0x00[1],
also the same for openbsd.
[1] https://github.com/torvalds/linux/blob/master/lib/string.c#L319.

The current U-Boot strlcat() contains trailing 0x00, I think it needs
to be updated.

Thanks,
Masahisa Kojima

>
> Best regards
>
> Heinrich
>
> > +
> > +     if (dlen >= size)
> > +             return ret;
> > +
> > +     dest += dstrlen;
> > +     size -= dlen;
> > +     if (len >= size)
> > +             len = size - sizeof(u16);
> > +
> > +     memcpy(dest, src, len);
> > +     dest[len >> 1] = u'\0';
> > +
> > +     return ret;
> > +}
> > +
> >   /* Convert UTF-16 to UTF-8.  */
> >   uint8_t *utf16_to_utf8(uint8_t *dest, const uint16_t *src, size_t size)
> >   {
>