Re: [PATCH 1/2] introduce "banned function" list

From: Eric Sunshine <sunshine@sunshineco.com>
Date: Jul 19, 2018, 21:59 UTC
Message-ID: <CAPig+cTbvKZ6Ui_ib2g2Xqbv8zFwJp9idocbWaZLxj0Lpe+YBA@mail.gmail.com>
In-Reply-To: <20180719212700.GA13151@sigill.intra.peff.net>

On Thu, Jul 19, 2018 at 5:27 PM Jeff King <peff@peff.net> wrote:

Show 16 quoted lines

> On Thu, Jul 19, 2018 at 05:11:15PM -0400, Eric Sunshine wrote:
> > On Thu, Jul 19, 2018 at 4:39 PM Jeff King <peff@peff.net> wrote:
> > > + * This header lists functions that have been banned from our code base,
> > > + * because they're too easy to misuse (and even if used correctly,
> > > + * complicate audits). Including this header turns them into compile-time
> > > + * errors.
> >
> > When the above talks about "including this header", the implication is
> > that it must be included _after_ the system header(s) which declare
> > the banned functions. I wonder if that requirement should be stated
> > here explicitly.
>
> Hmm, does it need to be? I had originally intended it to be included
> before, actually, though in the end I put it later.
> I guess it would yield declarations like strcpy_is_banned(), which would
> cause _different_ errors (probably link-time ones).

Yes, that's what I meant. You'd only get link-time errors if banned.h was included before the system headers (assuming I'm thinking about this correctly).

> > (Probably not worth a re-roll.)
>
> Yeah, I doubt it matters much either way, since the inclusion is done
> automatically in git-compat-util.h.

Exactly.

> I had also originally imagined this to be triggered via DEVELOPER=1,
> with something like "-include banned.h" in CFLAGS. But I think it
> probably is appropriate for everybody to run it, since it shouldn't
> cause any false positives or other compilation issues.

Agreed.

> The one I brainstormed (but forgot to mention) is that it might be
> possible for a platform to have strcpy as a macro already? In which case
> we'd need to #undef it or risk a compilation error (even if the macro
> isn't actually used).

I have some recollection (perhaps long outdated or just wrong) of Microsoft headers spewing deprecation warnings about "unsafe" functions. I don't know whether they did that by covering functions with macros or by decorating the function with a deprecation attribute or by some other mechanism, but such concern seems well-founded. #undef'ing them might indeed be a very good preventative tactic.