Let's not forget that the patch had been posted on the OpenSSL mailing list and had received a go ahead comment before that.
Having said that, if you're asking if there's a penetration test team that reviews all the patches. No there isn't. Like there isn't any such thing on 99.999999999% of all software that exists.
Last I knew Debian didn't do dedicated security review of patches to security-critical software, which is normal practice for other distributions.
No it wasn't. It was reading (and xoring into the randomness that would become the key being generated) uninitialised char values from an array whose address was taken, that results in unspecified values not undefined behaviour.
No, they XORed data from a bunch of entropy sources into an intermediate buffer (that was never initialised, because the whole point of it was to be random) and then XORed that into a buffer from which the key was made. Debian's patch removed that final XOR. It wasn't a bug in the original code (other than being hard to understand).
Can you please read about the issue before commenting more?
No it wasn't. The patch removed the read of the randomness buffer that folded it into another buffer (the MD_Update calls) because Valgrind was warning that the buffer it was reading from had never been initialised.
> Can you please read about the issue before commenting more?
Right back at you.
(meanwhile, long before this incident fedora just compiled openssl with -DPURIFY which disabled the bad behavior in a safe and correct way).
Then you run the tests, and if they pass, you package and upload it.
This allows a patch(set) can be sent to the upstream as a package saying "we did this, and if you want to include them, this apply cleanly to version x.y.z, any feedback is welcome".
Patches are maintained separately because debian doesn't normally repack the .tar.gz (or whatever) that the projects publish, as to not invalidate signatures and let people check that the file is in fact the same. An exception is done when the project publishes a file that contains files that cannot legally be redistributed.
OpenSSL already had an option to safely disable the bad behavior, -DPURIFY.
* https://udd.debian.org/patches.cgi?src=gnupg2&version=2.4.7-...
There is a lot of political stuff in there related to standards. For a specific example see:
* https://sources.debian.org/src/gnupg2/2.4.7-19/debian/patche...
The upstream GnuPG project (and the standards faction they belong to) specifically opposes the use of keys without user IDs as it is a potential security issue. It is also specifically disallowed by the RFC4880 OpenPGP standard. By working through the Debian process, the proponents of such keys are bypassing the position of the upstream project and the standard.
To be fair, in Debian's case politics come with the territory. Debian is a vision of what an OS should be like. With policies, standards & guidelines aimed at that, treating the OS as a whole.
That goes well beyond "gather packages, glue together & upload".
Same goes for other distributions I suppose (some more, some less).
Why? Heartbleed.
as always: imho (!)
i remember this incident - if my memory doesn't trick me:
it was openssl which accessed memory it didn't allocated to collect randomness / entropy for key-generation.
and valgrind complained about a possible memory-leak - its a profiling-tool with the focus on detecting memory-mgmt problems.
instead of taking a closer look / trying to understand what exactly went on there / causes the problem, the maintainer simply commented out / disabled those accesses...
mistakes happen, but the debian-community handled this problem very well - as in my impression they always do and did.
idk ... i prefere the open and community-driven approach from debian anytime over distributions which are associated to companies.
last but not least, the have a social contract.
long story short: at least for me this was an argument for the debian gnu/linux distribution, not against :))
just my 0.02€
It’s doubly important to upstream issues for security libraries: There are numerous examples of bad actors intentionally sabotaging crypto implementations. They always make it look like an honest mistake.
For all we know, prior or future debian maintainers of that package are working for some three letter agency. Such changes should be against debian policy.
[1] https://en.wikipedia.org/wiki/OpenSSL#Predictable_private_ke...