Re: [PATCH] string: Add tests for zero length string inputs

[Florian Weimer <fweimer at redhat dot com>]

On 09/23/2014 02:57 PM, Richard Earnshaw wrote:
> On 23/09/14 13:41, Florian Weimer wrote:
> > On 09/22/2014 06:09 PM, Richard Earnshaw wrote:
> > > > These tests are not testing null pointers, they are testing that when
> > > > given a zero length the functions actually read/write zero bytes.
> > > > Whether the specification demands that behaviour is arguable but I
> > > > believe that it is the most sane behaviour.
> >
> > > Valid pointers is more than just non-NULL.  In particular, it implies
> > > that is safe to dereference the addressed byte in a source operand even
> > > when the length parameter is zero.
> >
> > Valid pointers can also point one element past the end of an array of
> > objects.
>
> I don't think such a pointer forms a valid argument for a library
> function though.  See my previous reply to Paul.

They are a fairly common occurrence with the [first, last) iterator 
ranges in C++.  It's common to compute a pointer/length pair {first, 
last - first} and pass that to C functions, including C library functions.

This pattern is already incorrect in important corner cases (e.g., 
iterators derived from empty vectors), but do we really have to make the 
situation even worse?

--
Florian Weimer / Red Hat Product Security