I'm sorry for taking so long to get back to you.  After discussing with fellow 
devs, we'd prefer not to open a separate CVE for each item.  In looking at the 
items you helpfully gathered, we can categorize by type of problem and file 
formats affected.  I don't think we need to open a CVE for NPE or other parse 
exceptions (61286, 61287, 61059, pull53).  For the others, we could open a 
single CVE based on the poi-release (hey, these are now fixed in version 3.17) 
or we might open two -- one for permanent hangs, one for OOM?  My preference 
would be one CVE based on POI release.  

A full description in that one CVE will allow users to determine if 3.17 would 
protect them based on file type -- your main goal, right?

To fellow Devs and David, how does this sound?

DETAILS:

This is my understanding, please let me know if I've missed any or 
misunderstood the impacts.

61338 permanent hang : WMF
61295 OOM :doc, ppt, xls
61294 permanent hang : macros, wmf, emf, msg
52372 OOM: doc, ppt, xls

61286, 61287, 61059, pull 53 -- not an OOM or permahang

-----Original Message-----
From: [email protected] [mailto:[email protected]] 
Sent: Tuesday, September 19, 2017 2:44 PM
To: [email protected]
Subject: Re: RE: [ANNOUNCE] Apache POI 3.17 released

On 2017-09-19 07:56, "Allison, Timothy B." <[email protected]> wrote: 
> David,
>   Thank you for raising this issue.  If fellow devs are +1, I can fill out 
> the paper work.  Single CVE or multiple?
> 

My suggestion would be one CVE for each issue.  That way if a consuming project 
isn't affected by a particular vulnerability (e.g. the vulnerabilities affect a 
file type that the consumer doesn't use) they can avoid upgrading right away.

I believe the following are all vulnerabilities listed in the change log as 
being fixed since 3.16:

- 61338, "Avoid infinite loop in corrupt wmf"
- 61295, "Vector.read -- Java heap space on corrupt file"
- 61300, "Very slow processing on corrupted file"
- 61286, "can not deal with WriteProtectRecord element"
- 61287, "HeaderRecord or FooterRecord throws RecordFormatException when the 
text of length 0"
- 61294, "IOUtils.skipFully can run into infinite loop"
- 61059, "Fix incorrect use of short when unsigned short was required in 
NamePtg"
- pull 53, "Adding Null Pointer check"
- 52372, "OutOfMemoryError parsing a word file"

The good news is that all of these are denial of service vulnerabilities, which 
aren't too serious.

---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected] For additional 
commands, e-mail: [email protected]



---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to