Re: [ruby-core:63604] [ruby-trunk - Bug #10019] [Open] segmentation fault/buffer overrun in pack.c (encodes)

[Ramon de C Valle <rdecvalle () vmware com>]

On Jul 15, 2014, at 2:09 AM, cve-assign () mitre org wrote:

> Signed PGP part
> > Is MITRE or Red Hat going to assign a CVE for it?
>
> We haven't yet been able to determine whether the discussion is about
> two separate vulnerabilities.
>
> http://openwall.com/lists/oss-security/2014/07/09/13 says:
>
>   ruby -v: ruby 2.1.2p168 (2014-07-06 revision 46721) [i386-mingw32]
>
>   ...
>
>   While working with an AWS sample I hit a segmentation fault. The
>   same sample works under 1.9.3.
>
> First, we don't know what "The same sample works under 1.9.3" means.
> It might mean "The same AWS sample is also a working vulnerability
> reproducer when using Ruby 1.9.3." It might instead mean "With this
> AWS sample, my program works normally when using Ruby 1.9.3; in other
> words, no vulnerability is observed.”
It meant that his sample worked normally when he used Ruby 1.9.3. (I assumed this because the version he specified as 
containing the bug in the report was Ruby 2.1, and specified Ruby 2.0 as requiring backport, but not Ruby 1.9.3.)

> http://openwall.com/lists/oss-security/2014/07/10/15 says:
>
>   Anyway, whatever the reporter is referring to, he mentions it
>   doesn't occur in 1.9.3, and looking at 1.9.3, the only related
>   differences I immediately noticed are the absence of the check at
>   https://github.com/ruby/ruby/blob/trunk/pack.c#L829 in pack_pack
>   function and padding being an int (instead of char) in the encodes
>   function.
>
> These differences in pack.c obviously aren't the same as (and
> aren't expected to be the same as) the pack.c code changes in
> Revision 46778 (aka the
> https://bugs.ruby-lang.org/projects/ruby-trunk/repository/revisions/46778/diff/pack.c
> changes).
These are the differences I noticed when comparing the related code in Ruby 1.9.3 (where his sample worked normally) 
with Ruby 2.0 and 2.1, but are not related to the (off-by-one) initially discussed, as it may be the case the reporter 
may be referring to a different issue, since Tomas analysis of aws-sdk gem and its dependencies indicates it's unlikely 
this issue is being caused by any of these gems.

> (We realize that 1.9.3 is of interest because it is the "Old stable"
> distribution advertised on the
> https://www.ruby-lang.org/en/downloads/ page.)
>
> Is one of these scenarios the correct interpretation?
>
>   1. There is only one vulnerability. Version 2.1.2 is an example of
>      an affected version. Version 1.9.3 is an example of a
>      non-affected version.
>
>   2. There is only one vulnerability. Version 2.1.2 is an example of
>      an affected version. Version 1.9.3 is also an example of an
>      affected version.
>
>   3. A vulnerability in pack.c was fixed during Ruby 1.x development,
>      but then a regression occurred during Ruby 2.x development, and
>      the vulnerability is present in, for example, version 2.1.2.
>      (A regression would generally mean that two CVE IDs are
>      required.)
>
>   4. The Ruby 1.x pack.c and the Ruby 2.x pack.c are vulnerable in
>      substantially different ways, requiring different fixes.
>      (Again, this would generally mean that two CVE IDs are
>      required.)
>
> We don't require that the set of affected versions is precisely
> determined before a CVE assignment. Narrowing it down to one of the
> above scenarios is probably required because otherwise the correct
> number of CVE IDs isn't known.
Ruby 1.9.3, 2.0, and 2.1 are affected by the off-by-one. We’re still not sure about the presence of a different issue 
affecting Ruby 2.0 and 2.1. I left a comment on the report pointing out that 1.9.3 is also affected by the off-by-one 
and suggesting confirming with the reporter if he continues to observe the crash after adding the fix.

> --
> CVE assignment team, MITRE CVE Numbering Authority
> M/S M300
> 202 Burlington Road, Bedford, MA 01730 USA
> [ PGP key available through http://cve.mitre.org/cve/request_id.html ]
--
Ramon de C Valle
VMware Product Security Engineering

Attachment: signature.asc
Description: Message signed with OpenPGP using GPGMail