linux-kernel - Re: [patch] direct-io: propagate -ENOSPC errors

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] [day] [month] [year] [list]

Message-ID: <07bce0cc-efe2-9a73-09a4-3c041bf6067f@oracle.com>
Date:	Thu, 21 Apr 2016 10:17:50 -0400
From:	Todd Vierling <todd.vierling@...cle.com>
To:	Jeff Moyer <jmoyer@...hat.com>,
	Christoph Hellwig <hch@...radead.org>
Cc:	Alexander Viro <viro@...iv.linux.org.uk>,
	linux-fsdevel@...r.kernel.org, linux-kernel@...r.kernel.org
Subject: Re: [patch] direct-io: propagate -ENOSPC errors

On 03/21/2016 04:22 PM, Jeff Moyer wrote:
>> Just propagating some errors defintively seems odd.
> 
> Not really.  read, write, etc only expect a subset of errnos to be
> returned.  The goal was not to leak kernel-internal or unexpected error
> numbers to userspace, and I didn't think I would be able to successfully
> audit all code paths that lead here.  So, I opted for a more
> conservative patch that just allows one more errno through.

We have a use case which makes heavy use of dm to map plain files into
block devs. Propagating ENOSPC is useful here, as a sparse backing file
might run into this when extending, and the downstream app would be able
to see something other than EIO.

I'm ambivalent on whether or not to allow all errnos through, or just a
peephole like ENOSPC here. However, there's another errno which is
effectively the same cause as ENOSPC but triggered by a logical rather
than physical limit: EDQUOT.

If we're looking at a peephole for ENOSPC to be let through, I'd suggest
also allowing EDQUOT through. To your concerns about confusing
applications with too many possible errnos, EDQUOT could be translated
to ENOSPC before being let through to the block dev layer. (The net
effect is the same in any case, as the backing store has denied a write
due to some kind of space limit.)

-- 
-- Todd Vierling <tv@....org> <todd.vierling@...cle.com>