Bruno Haible <[EMAIL PROTECTED]> wrote:
> Jim Meyering wrote:
>> I noticed the following fix in glibc, so applied it here.
>
> Thanks. I find the ChangeLog entry confusing, since the problem is not with
> readlink, but with the terminating NUL byte that is added after the readlink
> call. So, we don't need to write a readlink() override - readlink() by itself
> is fine.
...
> !     * lib/canonicalize-lgpl.c (__realpath): Avoid buffer overflow after
> !     readlink on platforms without PATH_MAX.

Thanks.  I prefer your description, too.


Reply via email to