From: Sami Liedes Subject: Re: Intentionally corrupted vfat fs causing BUG Date: Sun, 12 Oct 2014 23:40:45 +0300 Message-ID: <20141012204045.GA12577@sli.dy.fi> References: <20141010205706.GJ27150@sli.dy.fi> <87h9z97aoh.fsf@devron.myhome.or.jp> Mime-Version: 1.0 Content-Type: multipart/signed; micalg=pgp-sha512; protocol="application/pgp-signature"; boundary="PEIAKu/WMn1b1Hv9" Cc: OGAWA Hirofumi , linux-fsdevel To: Richard Weinberger Return-path: Received: from gw01.mail.saunalahti.fi ([195.197.172.115]:51376 "EHLO gw01.mail.saunalahti.fi" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752701AbaJLUky (ORCPT ); Sun, 12 Oct 2014 16:40:54 -0400 Content-Disposition: inline In-Reply-To: Sender: linux-fsdevel-owner@vger.kernel.org List-ID: --PEIAKu/WMn1b1Hv9 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline On Sun, Oct 12, 2014 at 09:04:19PM +0200, Richard Weinberger wrote: > You misunderstood Sami's issue. He corrupted the vfat fs intentionally > to find issues > in the vfat driver. > And as he reports he found an nasty issue. > Any user can trigger a BUG_ON() using a crafted vfat image. > Please note, if you mount exactly the same image using msdos fs the issue > does not occur. Yeah, you can think of it as either a security issue if you wish, or just as a plain old robustness issue in the age of unreliable USB sticks etc. in that it just would be more ideal to fail gracefully instead of crashing. Anyway, I'm not advocating for any measure of severity (I leave that to others); I just find and report these in the hope that someone finds the reports useful. I personally view these more as robustness than security bugs at the moment, but certainly they can be seen as either. And if some of these get fixed, I will rerun the tests, so I might produce a daunting stream of reports. I know it would be nicer to report everything at once, but usually the issues found first are prevalent enough to mask other issues. By the way, I find it interesting that once I implemented a tool to minimize the differences between a bad fs and a good fs, every bug I have found in any filesystem implementation has minimized to a single bit flip. That suggests to me that fuzz testing is probably not very effective in finding bugs that require more than that. Sami --PEIAKu/WMn1b1Hv9 Content-Type: application/pgp-signature; name="signature.asc" Content-Description: Digital signature -----BEGIN PGP SIGNATURE----- Version: GnuPG v1 iQIcBAEBCgAGBQJUOudNAAoJEKLT589SE0a0n8kP/A0UxsBoIyQy+QbkiFkGIWVF OJ11D880b/2XPbA+eFpOOXexL13DA3r8DVaL9E+dlG96a6kq927hmx2exCeU05nn ag3WXjjfkQiBu5wmPqy42zZ3Yt89gioPwfxVwMe2yWJEiPLCNkman/DxmsO7NEvD R6mAWVXpeYdPSksdg2D+a5b8JZ1FVHF8sHHxqKkvjmeTlwwvhE5rtqfRyqr41mRu LW0i4PqxcneWFv/AzEmKPiArTydkKlivF1VKFm4ezqwWr2UgttcxL+qEtJ8iGQrY icxETwJtFxw/l0y8kazZMhM6Yy59sgrNCaYC+roFgV/TayyigEG2rgTsndFPse2U lG17Q5WQqpZdrvDoLHU/nj1cFRKl1rZnFHymqvPQ5Beavus32Yd3voLrjsi0hW2D m7lgI8TcPr5N+bD7VFt4J5swzEXNhWSosfHAxxxAwTmP5D99x00thnSkDCy3YRu4 9+6ym3n+SZ+VdnYu6kBVcXdlFrlPjEvLlVpDn64028Xkdt16SVO720bTIhZrLyT7 8rOsKlj4vFqWF3AuWAfWlR+A6DgFOiPfrotNgst4zhBSoiaClNGfzyLmQ6M6orb5 A9kf+FOvm3ggciUjpKldXFEJnOo4/u9/c17bxcrXTz/YV/PTqk+o42/mYkk03EiH bsoCd91oN1TNZVcY1qdd =u3X2 -----END PGP SIGNATURE----- --PEIAKu/WMn1b1Hv9--