[tpmoney]

> In addition to your point, it seems obvious that disclosure policy for FOSS should be “when patch available” and not static X days.

So when the xz backdoor was discovered, you think it would have been better to sit on that quietly and try to both wrest control of upstream away from the upstream maintainers and wait until all the downstream projects had reverted the changes in their copies before making that public? Personally I'm glad that went public early. Yes there is a tradeoff between speed of public disclosure and publicity for a vulnerability, but ultimately a vulnerability is a vulnerability and people are better off knowing there's a problem than hoping that only the good guys know about it. If a Debian bug starts tee-ing all my network traffic to the CCP and the NSA, I'd rather know about it before a patch is available, at least that way I can decide to shut down my Debian boxes.

[Orygin]

The XZ backdoor is not a bug but a malicious payload inserted by malicious actors. The security vulnerability would immediately been used as it was created by attackers.

This bug is almost certainly too obscure to be found and exploited in the time the fix can be produced by Ffmpeg. On the other hand, this vuln being public so soon means any attacker is now free to develop their exploit before a fix is available.

If Google's goal is security, this vulnerability should only be disclosed after it's fixed or a reasonable time (which, according to ffmpeg dev, 90 days is not enough because they receive too many reports by Google).

[tpmoney]

A bug is a bug, regardless of the intent of the insertion. You have no idea if this bug was or wasn't intentionally inserted. It's of course very likely that it wasn't, but you don't and can't know that, especially given that malicious bug insertion is going to be designed to look innocent and have plausible deniability. Likewise, you don't know that the use of the XZ backdoor was imminent. For all you know the intent was to let it sit for a release or two, maybe with an eye towards waiting for it to appear in a particular down stream target, or just to make it harder to identify the source. Yes, just like it is unlikely that the ffmpeg bug was intentional, it's also unlikely the xz backdoor was intended to be a sleeper vulnerability.

But ultimately that's my point. You as an individual do not know who else has access or information about the bug/vulnerability you have found, nor do you have any insight into how quickly they intend to exploit that if they do know about it. So the right thing to do when you find a vulnerability is to make it public so that people can begin mitigating it. Private disclosure periods exist because they recognize there is an inherent tradeoff and asymmetry in making the information public and having effective remediations. So the disclosure period attempts to strike a balance, taking the risk that the bug is known and being actively exploited for the benefit of closing the gap between public knowledge and remediation. But inherently it is a risk that the bug reporter and the project maintainers are forcing on other people, which is why the end goal must ALWAYS be public disclosure sooner rather than later.

[Orygin]

A 25 years old bug in software is not the same as a backdoor (not a bug, a full on backdoor..). The bug is so old if someone put it there intentionally, well congrats on the 25yo 0day.

Meanwhile the XZ backdoor was 100% meant to be used. I didn't say when and that doesn't matter, there is a malicious actor with the knowledge to exploit it. We can't say the same regarding the bug in a 1998 codec that was found by extensive fuzzing, and without obvious exploitation path.

Now, should it be patched? Absolutely, but should the patch be done asap at the cost of other maybe more important security patches? Maybe, maybe not. Not all bugs are security vulns, and not all security vulns are exploitable

[tpmoney]

> Absolutely, but should the patch be done asap at the cost of other maybe more important security patches? Maybe, maybe not. Not all bugs are security vulns, and not all security vulns are exploitable

I fully agree which is why I really don’t understand why everyone is all up in arms here. Google didn’t demand that this bug get fixed immediately. They didn’t demand that everything be dropped to fix a 25 year old bug. They filed a (very good and detailed) bug report to an open source product. They gave a private report out of courtesy and an acknowledgment of the tradeoffs inherent in public bug disclosure, but ultimately a bug is a bug, it’s already public because the source code is public. If the ffmpeg devs didn’t feel it was important to fix right away, nothing about filing a bug report, privately or publicly changes any of that.

[Orygin]

In the end, a report saying "fix this within 90 days or this gets public" for small-ish bugs like this is a kind of demand. Do this or this gets out and you'll have to make an express release to fix it anyway.

I can understand that stance for serious bugs and security vulnerabilities. I can understand such delays for a company with a big market cap to put pressure on them. But these delays are exactly like a demand put on the company: fix it asap or it gets public. We wouldn't have to do this if companies in general didn't need to get publicly pressured into fixing their stuff. Making it public has two objectives: Warn users they may be at risk, and force the publisher to produce a fix asap or else risk a reputation hit.

> If the ffmpeg devs didn’t feel it was important to fix right away, nothing about filing a bug report, privately or publicly changes any of that.

It does change how they report. Had they given more time or staggered their reports over time, Ffmpeg wouldn't have felt pressure to publish fixes asap. Even if the devs can say they won't fix, any public project will want to keep a certain quality level and not let security vulnerabilities get public.

In the end, had these reports been made by random security researchers, no drama would have happened. But if I see Google digging up 25 years old bugs, is it that much to expect them to provide a patch with it?

[jsnell]

But this isn't a "small-ish bug". What gave you that impression? It's a vulnerability in code that is both compiled in by default, and that is reachable when ffmpeg is run with its default settings when run on a file crafted to trigger the bug.

And if you believe this is a "small-ish" bug just because the ffmpeg Twitter account's gaslighting about "20 frames of a single video in Rebel Assault", then surely it being disclosed would be irrelevant? The only way the disclosure timeline makes a difference is if ffmpeg too think that the bug is serious.

[tpmoney]

> In the end, a report saying "fix this within 90 days or this gets public" for small-ish bugs like this is a kind of demand. Do this or this gets out and you'll have to make an express release to fix it anyway.

I think this is where the disconnect is. To my mind there is no "do this or else" message here, because there is no "or else". The report is a courtesy advance notice of a bug report that WILL be filed, no matter what the ffmpeg developers do. It's not like this is some awful secret that Google is promising not to disclose if ffmpeg jumps to their tune.

Further, the reality is most bug reports are never going to be given a 90 day window. Their site requests that if you find a security vulnerability you email their security team, but it doesn't tell you not to also file a bug report, and their bug report page doesn't tell you not to file anything you think might be a security or vulnerability bug to the tracker. And a search through the bug tracker shows more than a few open issues (sometimes years old) reporting segfault crashes, memory leaks, un-initialized variable access, heap corruption, divide by zero crashes, buffer overflows, null pointer dereferences and other such potential safety issues. It seems the ffmpeg team has no problems generally with having a backlog of these issues, so certainly one more in a (as we've been repeatedly reminded) 25 year old obscure codec parser is hardly going to tank their reputation right?

> In the end, had these reports been made by random security researchers, no drama would have happened.

And now we get to what is really the heart of the matter. If anyone else has reported this bug in this way, no one would care. It's not that Google did anything wrong, it's that Google has money so everyone is mad that they didn't do even more than they already do. And frankly that attitude stinks. It's hard enough getting corporations to actually contribute back to open source projects, especially when the license doesn't obligate them to at all. I'm not advocating holding corporations to some lesser standard, if the complaint was that Google was shoving unvalidated, and un-validatable low effort reports into the bug tracker, or that they actually were harassing the ffmpeg developers with constant followups on their tickets and demands for status updates then that would be poor behavior that we would be equally upset about if it came from anyone. But like you said, any other security researcher behaving the same way would be just fine. Shitting on Google this way for behaving according to the same standards outlined on ffmpeg's own website because of who they are and not what they've done just tells other corporations that it doesn't matter if you contribute code and money in addition to bug reports, if you don't do something to someone's arbitrary standard based on WHO you are, rather than WHAT you do, you'll get shit on for it. And that's not going to encourage more cooperation and contributions from the corporations that benefit from these projects.