linux-kernel - Re: [BUG] libmount misparses mountinfo on Linux v2.6.39-rc1

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]

Message-Id: <E1Q5bLG-0001eG-9d@pomaz-ex.szeredi.hu>
Date:	Fri, 01 Apr 2011 12:12:34 +0200
From:	Miklos Szeredi <miklos@...redi.hu>
To:	Karel Zak <kzak@...hat.com>
CC:	miklos@...redi.hu, aneesh.kumar@...ux.vnet.ibm.com,
	victor.vde@...il.com, util-linux@...r.kernel.org,
	linuxram@...ibm.com, viro@...iv.linux.org.uk, ejmarkow@...oo.com,
	linux-fsdevel@...r.kernel.org, linux-kernel@...r.kernel.org,
	lennart@...ttering.net
Subject: Re: [BUG] libmount misparses mountinfo on Linux v2.6.39-rc1

On Fri, 1 Apr 2011, Karel Zak wrote:
> On Fri, Apr 01, 2011 at 10:45:48AM +0200, Miklos Szeredi wrote:
> > On Fri, 01 Apr 2011, Aneesh Kumar K. V wrote:
> > > On Thu, 31 Mar 2011 22:24:01 +0200, victor.vde@...il.com wrote:
> > > > /proc/self/mountinfo on Linux v2.6.39-rc1 includes the UUID of ext4
> > > > and ext3 mounts, for example:
> > > > 
> > > > 15 1 8:3 / / rw,noatime uuid:c645234d-9756-4d84-825e-6fe999252a34 - ext4 /dev/sda3 rw,user_xattr,acl,barrier=1,data=ordered
> > > > 
> > > > It seems the hyphens in the uuid confuse the parser
> > > > mnt_parse_mountinfo_line in tab_parse.c of libmount in util-linux.
> > > 
> > > shouldn't the parser look for " " followed by "-" followed by " " as the
> > > field seperator ? 
> > > 
> > > I am adding Miklos and Ram Pai to CC to check whether the kernel or the libmount 
> > > should be fixed ?
> > 
> > I think libmount should be fixed, though there is always the question
> > of backward compatibility.
> 
>  The parser uses
> 
> 	rc = sscanf(s,	"%u "		/* (1) id */
> 			"%u "		/* (2) parent */
> 			"%u:%u "	/* (3) maj:min */
> 			"%ms "		/* (4) mountroot */
> 			"%ms "		/* (5) target */
> 			"%ms"		/* (6) vfs options (fs-independent) */
> 			"%*[^-]"	/* (7) optional fields */

So this says, "skip everything up to the first hyphen", even though
the rule might be better expressed as "skip all space delimited words
up to the first standalone hyphen".  I guess that's not possible to
express with a scanf expression, though.

> 			"- "		/* (8) separator */
> 			"%ms "		/* (9) FS type */
> 			"%ms "		/* (10) source */
> 			"%ms",		/* (11) fs options (fs specific) */
> 
>  note that almost the same code is in systemd.
> 
> > However I don't see how the kernel could be fixed, given that libmount
> > doesn't seem to parse escape sequences (e.g. "\040" for space), which
> > it also should.
> 
>  It calls unmangle_string() for all fields, so all sequences should be
>  decoded.

Okay, so the kernel can simply escape the troublesome hyphens, which
would make the human parsing of /proc/pid/mountinfo more difficult but
at least would fix compatibility with util-linux's parser.

Thanks,
Miklos
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/