Re: [PATCH 2/2] push -s: skeleton

From: Jeff King <peff@peff.net>
Date: Sep 9, 2011, 19:12 UTC
Message-ID: <20110909191240.GA30019@sigill.intra.peff.net>
In-Reply-To: <20110909160301.GA9707@gnu.kitenet.net>

On Fri, Sep 09, 2011 at 12:03:01PM -0400, Joey Hess wrote:

> The most credible attack I have so far does not involve binary files in
> tree. Someone pointed out that git log, git show, etc stop printing
> commit messages at NULL.

It was me.

> It might be worth ameloriating that attack by making git log always
> show the full buffer. Or it would be easy to write a tool that finds
> any commits that have a NULL in their message.

Unfortunately, that is going to involve a pretty huge code audit of git, as the "tack a \0 to the end of an object just in case" code dates back quite a while (e871b64, unpack_sha1_file: zero-pad the unpacked object, 2005-05-25). So I suspect there is a lot of code built on top of the assumption that commit messages are NUL-terminated strings.

Besides which, that is only one form of hiding. If collision attacks against sha1 become a possibility, I think we are better to talk about moving to a new hash. Even sha-256 truncated to 160 bits would be better than sha-1 (AFAIK, that family of SHA does not suffer from the same attacks, so we would still be in the 2^80 range for collision attacks).

-Peff