Skip to main content

in reply to daniel:// stenberg://

sucks that you have to deal with this :(
Do you have any idea who is filing these and what their motivation is? Or do they believe these to be genuine bugs?
in reply to Ondřej Surý

@ondrej it happened before we became CNA but yeah, there is clearly no filtering what so ever
in reply to daniel:// stenberg://

MITRE page contains a little bit more information: cve.mitre.org/cgi-bin/cvename.…

Hope this bullshit will not cease a little bit once you became your own CNA. There should be some penalty for people filling bogus CVEs with MITRE…

in reply to daniel:// stenberg://

@hanno I thought the purpose of becoming a CNA was so they would send these requests to you instead of blindly filing them
in reply to Josh Bressers

@joshbressers @hanno correct, but this was filed before we were accepted. We are now getting it assigned so I believe we will get the mandate to reject it... we'll soon find out!
in reply to daniel:// stenberg://

@joshbressers @hanno

AFAIR, to reject it has a very high barrier, like not being relevant to the project it's assigned to at all. Or pointing at features/code paths not available and such things. Basically the content of CVE record is completely wrong, not the claim towards the project itself.

I would expect DISPUTED being the right solution, preferably with a URL to the detail why it's wrong.

There is a closed CVE slack where such things can be discussed, though.

in reply to daniel:// stenberg://

I like the "Thanks a lot!" at the bottom. Both thanks for reading this and thanks for making me feel with this. 😜
in reply to daniel:// stenberg://

> The curl security team will work on getting this CVE rejected.

Good luck on that. At best, it will be labelled as "disputed".

I would recommend you to become a CNA yourself for curl. The general CVE policy is that any reporter need to first reach out to the appropriate CNA to get a CVE for an issue. A reporter can go to a root CNA too, but only after the project CNA has been reached. And there are some conflict resolution policy if the CNA and reporter disagrees.

It's a bit of "paperwork" and administrative steps to get approved as a CNA. You need to document security policies and processes and show you understand how CVE's are worded and registered. But I would expect curl to be in a good position to get approved.

We did that a while back in OpenVPN. And the tooling (in particular cvelib) makes it pretty easy to register and publish CVEs.

in reply to daniel:// stenberg://

This is perfect...

"We cannot say which and the distinction does not matter to us."

in reply to daniel:// stenberg://

"Do not blindly trust the CVE system. It is full of cracks and bogus reports such as CVE-2023-52071."

Okay that seems a bit extreme. Sure do your research seems verify stuff but is it really full of bogus reports?

in reply to 0ddj0bb

@0ddj0bb yes it is. Numerous projects get the same kind of crap. This is not an isolated case. I can't say numbers, but I would say there are easily hundreds of equally bad/bogus ones.
in reply to daniel:// stenberg://

i wouldn't call 5.8% of cves to be "full" of junk.

Maybe I'm just focusing too much on full there. Just seems unnecessary hyperbole that tries to call the entire nvd into question on its credibility.

It does seem nvd does respond well to rejection process.

in reply to 0ddj0bb

@0ddj0bb NVD does not reject things at all. If MITRE does not reject a CVE, NVD will host it. Been there, tried that.
in reply to daniel:// stenberg://

they've rejected over 13000 cves. So i can't find that they don't reject anything.
in reply to 0ddj0bb

that says there are 13,000 rejected ones. Not that NVD rejected them.
This entry was edited (9 months ago)
in reply to daniel:// stenberg://

AFAIK this is the second totally bogus CVE for curl! I'm not happy with the people who are doing this. This needs to be stopped ASAP
in reply to daniel:// stenberg://

Wait, what? I can get up to 5 kilodollars for reporting bugs that curl has already fixed as new security issues and providing the patch, that curl has already made, as a fix? Where do I sign up? I could start buying LEGO if this works.
in reply to daniel:// stenberg://

I like how the description is just straight up wrong:

"The product uses untrusted input when calculating or using an array index, [...]",

when the patch they even refer to makes it obvious that there is no input involved at all.

in reply to daniel:// stenberg://

"Hey David, the security of our users is our top priority! Please tell me in all details how you plan to fix this vulnerability in our systems."