Bug#695224: Locale::Maketext security fix: real world breakage?

2013-02-06 Thread Giuseppe Iuculano
Hi Dominic,

On 04/02/2013 21:28, Dominic Hargreaves wrote:
 I had no replies about this, so I think it's time to bite the bullet
 and decide whether we should target this fix at
 
 - stable-security
 - stable
 - neither of the above.
 
 I think I'm leaning towards stable on the basis that that's a slightly
 safer place to land a possibly-problematic fix, as well as the fact I
 don't know of any real world exploits for this, but I an open to (and
 welcome) all comments.
 
 I seem to remember reading that a point release of squeeze is
 due quite soon, but I couldn't find an announcment of such.

from http://openwall.com/lists/oss-security/2012/12/11/4:

I think the vulnerability is effective only when attacker has first
argument of maketext() under control.

However that means the attacker can run any code even without this
`vulnerability'. It's like saying glibc's gettext() is vulnerable. But
that's not true.

Sure gettext(%s, user_input) is not safe, but this is flaw in the
caller, not in the gettext. The same applies to
Locale::Maketext::maketext().

Petr Pisar 2012-12-06 11:18:46 EST


This is CVE-2012-6329 and I think this doesn't warrant a DSA, please fix
it in stable.


Cheers,
Giuseppe.



signature.asc
Description: OpenPGP digital signature


Bug#695224: Locale::Maketext security fix: real world breakage?

2013-02-04 Thread Dominic Hargreaves
On Fri, Jan 18, 2013 at 03:06:38PM +, Dominic Hargreaves wrote:
 On Wed, Dec 05, 2012 at 04:05:01PM -0500, Ricardo Signes wrote:
  * Dominic Hargreaves d...@earth.li [2012-12-05T13:51:19]
   I wondered (and the question has arised within the Debian project) whether
   anyone might be relying on the previous behaviour? Have you been able to 
   do
   any assessment of this?
  
  It's difficult to say, unfortunately, because (I suppose) most projects that
  would use Locale::Maketext would not be CPAN projects, and so finding them 
  is
  not trivial.
  
  I did do some grepping of the CPAN and found zero cases.
  
  It should be quite easy to add this behavior back as optional, if we find
  we've broken anything.
 
 Hi,
 
 A fix for that has been in Debian unstable/testing for the past month
 and we've had no reports of problems. That doesn't mean everything, of
 course, but it is probably time to decide whether to push this out to
 Debian stable. As such I'd be very interested in hearing from anyone
 who has real world examples of this breaking things.

I had no replies about this, so I think it's time to bite the bullet
and decide whether we should target this fix at

- stable-security
- stable
- neither of the above.

I think I'm leaning towards stable on the basis that that's a slightly
safer place to land a possibly-problematic fix, as well as the fact I
don't know of any real world exploits for this, but I an open to (and
welcome) all comments.

I seem to remember reading that a point release of squeeze is
due quite soon, but I couldn't find an announcment of such.

Dominic.

-- 
Dominic Hargreaves | http://www.larted.org.uk/~dom/
PGP key 5178E2A5 from the.earth.li (keyserver,web,email)


-- 
To UNSUBSCRIBE, email to debian-bugs-rc-requ...@lists.debian.org
with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org



Bug#695224: Locale::Maketext security fix: real world breakage?

2013-01-18 Thread Dominic Hargreaves
On Wed, Dec 05, 2012 at 04:05:01PM -0500, Ricardo Signes wrote:
 * Dominic Hargreaves d...@earth.li [2012-12-05T13:51:19]
  I wondered (and the question has arised within the Debian project) whether
  anyone might be relying on the previous behaviour? Have you been able to do
  any assessment of this?
 
 It's difficult to say, unfortunately, because (I suppose) most projects that
 would use Locale::Maketext would not be CPAN projects, and so finding them is
 not trivial.
 
 I did do some grepping of the CPAN and found zero cases.
 
 It should be quite easy to add this behavior back as optional, if we find
 we've broken anything.

Hi,

A fix for that has been in Debian unstable/testing for the past month
and we've had no reports of problems. That doesn't mean everything, of
course, but it is probably time to decide whether to push this out to
Debian stable. As such I'd be very interested in hearing from anyone
who has real world examples of this breaking things.

Cheers,
Dominic.

-- 
Dominic Hargreaves | http://www.larted.org.uk/~dom/
PGP key 5178E2A5 from the.earth.li (keyserver,web,email)


-- 
To UNSUBSCRIBE, email to debian-bugs-rc-requ...@lists.debian.org
with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org