linux-kernel - Re: [PATCH v3 05/15] coda: remove CODA

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:   Thu, 10 Nov 2022 22:13:32 -0500
From:   Jan Harkes <jaharkes@...cmu.edu>
To:     Ahelenia Ziemiańska 
        <nabijaczleweli@...ijaczleweli.xyz>
Cc:     Jonathan Corbet <corbet@....net>,
        Federico Vaga <federico.vaga@...a.pv.it>,
        Alex Shi <alexs@...nel.org>,
        Yanteng Si <siyanteng@...ngson.cn>,
        Hu Haowen <src.res@...il.cn>,
        Greg Kroah-Hartman <gregkh@...uxfoundation.org>,
        Jiri Slaby <jirislaby@...nel.org>,
        Jakub Kicinski <kuba@...nel.org>, linux-doc@...r.kernel.org,
        linux-kernel@...r.kernel.org,
        linux-doc-tw-discuss@...ts.sourceforge.net
Subject: Re: [PATCH v3 05/15] coda: remove CODA_MAGIC

On Thu, Nov 10, 2022 at 08:14:01PM -0500, Ahelenia Ziemiańska wrote:
> We have largely moved away from this approach, and we have better
> debugging tooling nowadays: kill it.

I still haven't received a response to my earlier email asking what
better debug tooling you are talking about.

I gave an example of an older bug, which has been fixed, where a Coda
inode accidentally ending up in ext4 through an mmap path. This ended up
scribbling over a spinlock in the coda_inode_info data which resulted
into pretty random system lockups.

This was actually quite hard to track down because there is no check
that `file_inode(vma->vm_file)` actually returns the inode of another
file system. So I'd be interesting what this tooling is or how it can
be used to better catch/identify such cases.

So I actually like this type of magic because it can potentially catch
some cases that should never happen. We could rename it to from MAGIC to
ASSERT, we don't remove asserts from code because of better debug tooling.
Those random 4 bytes in a Coda internal structure are not in a hot path.

Jan