Re: [PATCH] Fix buffer not null terminated on (ecpg lib)

Hi,

On 2020-04-23 14:36:15 +0900, Kyotaro Horiguchi wrote:
> At Thu, 23 Apr 2020 01:21:21 -0300, Ranier Vilela <ranier.vf@gmail.com> wrote in
> > Em qua., 22 de abr. de 2020 às 23:27, Kyotaro Horiguchi <
> > horikyota.ntt@gmail.com> escreveu:
> > >
> > > -       strncpy(sqlca->sqlerrm.sqlerrmc, message,
> > > sizeof(sqlca->sqlerrm.sqlerrmc));
> > > -       sqlca->sqlerrm.sqlerrmc[sizeof(sqlca->sqlerrm.sqlerrmc) - 1] = 0;
> > > +       sqlca->sqlerrm.sqlerrmc[sizeof(sqlca->sqlerrm.sqlerrmc) - 1] =
> > > '\0';
> > > +       strncpy(sqlca->sqlerrm.sqlerrmc, message,
> > > sizeof(sqlca->sqlerrm.sqlerrmc) - 1);
> > >
> > > The existing strncpy then terminating by NUL works fine. I don't think
> > > there's any point in doing the reverse way.  Actually
> > > sizeof(sqlca->sqlerrm.sqlerrmc) - 1 is enough for the length but the
> > > existing code is not necessarily a bug.
> > >
> > Without understanding then, why Coveriy claims bug here.
>
> Well, handling non-terminated strings with str* functions are a sign
> of bug in most cases.  Coverity is very useful but false positives are
> annoying.  I wonder what if we attach Coverity annotations to such
> codes.

It might be worth doing something about this, for other reasons. We have
disabled -Wstringop-truncation in 716585235b1. But I've enabled it in my
debug build, because I find it useful. The only warning we're getting
in non-optimized builds is

/home/andres/src/postgresql/src/interfaces/ecpg/ecpglib/misc.c: In function ‘ECPGset_var’:
/home/andres/src/postgresql/src/interfaces/ecpg/ecpglib/misc.c:565:17: warning: ‘strncpy’ output truncated before terminating nul copying 5 bytes from a string of the same length [-Wstringop-truncation]
  565 |                 strncpy(sqlca->sqlstate, "YE001", sizeof(sqlca->sqlstate));
      |                 ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~