Re: [secdir] Secdir review of draft-murchison-nntp-compress-05

[Julien ÉLIE <[e-mail]>]

Hi Barry,

I'm currently reviewing the comments I need to take into account for the 
document.
Do you believe I should add a note about a possible security issue as 
far as the use of DEFLATE is concerned?  (see below)
(An "out of memory" attack?)

Thanks,

-- 
Julien

Le 22/09/2016 à 22:52, Julien ÉLIE a écrit :
> Hi Barry,
>
> Thanks for your answer.  I'll have a deeper look at it soon.
>
> As for the remaining "small points" of rewording (SHOULD, MUST, etc.),
> I'll calmly see them when updating the draft after Last Call.  In case I
> have a doubt for the best wording, I'll ask Ken (co-author) and Michael
> (write-up shepherd) their preference.  We'll obviously manage to come up
> with the "final" rewording as we'll have 3 votes for 2 choices.
>
>
> Now that I think about it, wouldn't it be worth adding something about
> possible "out of memory" attacks in the Security Considerations Section?
>
> According to <http://zlib.net/zlib_tech.html>, "a 50MB file filled with
> zeros [...] a version of run-length encoding optimized for this sort of
> unusual data file [...] could encode the test file in five bytes. That
> would be a compression factor of 10,000,000:1".
> So sending just 5 bytes may require an expansion to 50MB.
>
> Implementations therefore really need to check that the uncompressed
> data does not exceed an upper bound limit.
>
> RFC 1951 (DEFLATE) does not mention it in its Security Considerations,
> and I don't remember having seen that in other RFCs too.  Maybe we
> should say a word for it in the NNTP COMPRESS extension.
>