From: Alex Riesen <raa.lkml@gmail.com>
Date: Fri, 21 Jul 2006 14:54:57 GMT
Subject: Re: Unanticipated test error
Message-ID: <81b0412b0607210754m1e3c8bf9ne717786e666fa7e1@mail.gmail.com>
In-Reply-To: <20060721081954.GA29645@bohr.gbar.dtu.dk>

On 7/21/06, Peter Eriksen <s022018@student.dtu.dk> wrote:
> ...
> > Well, there are differences. Correct translation from memcpy
> > to strlcpy (aside the fact with \0 inside the string) would be
> > something like:
> >
> >  strlcpy(to, from, len + 1);
> >
> > assuming your example with memcpy. strlcpy expects size of
> > storage, and will never write more bytes that it was allowed to.
> > That'll cut off last character of the source string, unless it is
> > \0-terminated before the size of storage.
>
> I see it now.  What I did was wrong.  Appending " + 1" to every
> one of my calls makes the patch survive "make test".  However,
> since strlcpy() calls strlen(from), it would have to be checked,
> that 'from' is always NUL terminated.  The benefits of this patch
> seem to shrink.

Probably, but you still have room to balance benefits.