[ 
https://issues.apache.org/jira/browse/HADOOP-6966?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
 ]

Greg Roelofs updated HADOOP-6966:
---------------------------------

    Attachment: bzip2-1.0.6-CVE-2010-0405.patch.txt
                H-6966.v1.trunk-hadoop-common.patch.txt

Relevant portion of bzip2 1.0.6 patch for CVE-2010-0405, and untested (not even 
compiled) port of it to Hadoop trunk.

I've also added a credit that I suspect should have been added long ago (but in 
any case is relevant to this patch) and marked three of the missing 
bounds-check locations without actually attempting to fix them.  (They're all 
trivial, I think, just tedious; and I don't have time at the moment to finish 
this off.)

> missing bzip2 bounds checks, other stream-corruption checks
> -----------------------------------------------------------
>
>                 Key: HADOOP-6966
>                 URL: https://issues.apache.org/jira/browse/HADOOP-6966
>             Project: Hadoop Common
>          Issue Type: Bug
>    Affects Versions: 0.22.0
>            Reporter: Greg Roelofs
>         Attachments: bzip2-1.0.6-CVE-2010-0405.patch.txt, 
> H-6966.v1.trunk-hadoop-common.patch.txt
>
>
> The bzip2 Java port by Keiron Liddle appears to be very closely related to 
> the bzip2/libbzip2 C code, but it is missing a number of data-integrity 
> checks present in the latter (including today's fix for CVE-2010-0405).
> Java may be immune to C-style buffer overruns, but the data-corruption 
> possibility remains.  I believe most of the 20+ BZ_DATA_ERROR checks in 
> decompress.c are appropriate in the Ant/Hadoop port.

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.

Reply via email to