Re: [PATCH 1/2] [PATCH 1/2] Doc: deprecated.rst: add strlcat()

[Kees Cook <kees@kernel.org>]

On Mon, May 11, 2026 at 02:26:49PM +0100, David Laight wrote:
> On Mon, 11 May 2026 13:40:55 +0200
> Geert Uytterhoeven <geert@linux-m68k.org> wrote:
> 
> > Hi Manuel,
> > 
> > On Sun, 10 May 2026 at 18:52, Manuel Ebner <manuelebner@mailbox.org> wrote:
> > > add strlcat and alternatives  
> > 
> > Thanks for your patch!
> > 
> > > --- a/Documentation/process/deprecated.rst
> > > +++ b/Documentation/process/deprecated.rst
> > > @@ -162,6 +162,12 @@ if a source string is not NUL-terminated. The safe replacement is strscpy(),
> > >  though care must be given to any cases where the return value of strlcpy()
> > >  is used, since strscpy() will return negative errno values when it truncates.
> > >
> > > +strlcat()
> > > +---------
> > > +strlcat() must re-scan the destination string from the beginning on each
> > > +call (O(n^2) behavior). Alternatives are seq_buf_puts(), seq_buf_printf(),
> > > +snprintf() and scnprintf()  
> > 
> > The last two not only require the caller to keep track of the offset
> > in the buffer, but also using "%s" when storing passed strings.
> 
> Which also means they are significantly slower.
> Mind you, some code has:
> 	strlcat(buf, "\n", SIZE);
> 	return strlen(buf);
> which carefully scans the string twice.
> Since the '\0' isn't always needed (eg 'show' functions), this can be:
> 	len = strlen(buf);
> 	buf[len] ='\n';
> 	return len + 1;
> Of course, the code could often easily get the length by other means.

I think I'd prefer to only recommend using seq_buf API. Or for sysfs,
sysfs_emit() as seq_buf hasn't been extended there yet.

-Kees

-- 
Kees Cook