Re: [PATCH] xrealloc: do not reuse pointer freed by zero-length realloc()

From: Junio C Hamano <gitster@pobox.com>
Date: Sep 2, 2020, 19:19 UTC
Message-ID: <xmqqo8moasvp.fsf@gitster.c.googlers.com>
In-Reply-To: <20200902075439.GA855335@coredump.intra.peff.net>

Jeff King <peff@peff.net> writes:

Show 12 quoted lines

> There are other variants, too:
>
>   - we could use malloc(1) versus xmalloc(0). Maybe more
>     readable/obvious? But also potentially allocates an extra byte when
>     the platform malloc(0) would not need to.
>
>   - we could return a non-NULL "ptr" without shrinking it at all (nor
>     allocating anything new). This is perfectly legal, and the
>     underlying realloc() would still know the original size if anybody
>     ever asked to grow it back again.
>
> I have to admit I don't overly care between them.

I don't either. I admit that the latter I didn't think of---it feels tricky and harder to reason about than any other variants.

Show 7 quoted lines

> I suspect one of the
> reasons we never ran into this 15-year-old bug is that it's quite hard
> to convince Git to call realloc(0) in the first place. I only saw it
> when investigating a bug in another series, and there the problem turned
> out to be reading garbage bytes off the end of a buffer (which we
> interpreted as a serialized ewah bitmap which happened to have a zero in
> its length byte).

Thanks.