Live data from Hacker News

Technical Details on the Recent Firefox Add-On Outage

hacks.mozilla.org

271–279 of 279 posts

Re: Technical Details on the Recent Firefox Add-On Outage

#271
post #177
post #97

> users should be able to opt-in to updates (including hot-fixes) but opt out of everything else Finally some good news. This is what I suggested in one of the previous threads: there should be a delivery channel for important updates, and a channel for experiments/telemetry/whatnot. Some other HNer said it was an unrealistic expectation "because manpower". Guess what, it isn't. This is how things should always be.

They did have that capability (see https://wiki.mozilla.org/Firefox/Go_Faster/System_Add-ons/Pr... ), at least the linked repository had commits in 2016. I can't tell from a 30 second search why that no longer works, just that it is a replacement of a similar previous capability.

This started before my time at Mozilla, but I think several different systems grew out of the "go faster" initiative. Of them I believe the fastest and most flexible means of shipping changes is Normandy, which is primarily used for our shield studies. Like the blog post says, we need to revisit this.

(Disclosure: I work for Mozilla)

Re: Technical Details on the Recent Firefox Add-On Outage

#272
post #263

Earlier quoted context omitted.

I think that would be categorised as an option "that these greyware apps [could] change and plausibly claim they were acting on behalf of the user’s “choice” not to opt-out of the light grey checkbox on page 43 of their EULA". Sorry, but I don't really see how. We've been using click-to-play safeguards on embedded content for years, and they have proved highly effective at stopping abusive or outright malicious conte…

> Sorry, but I don't really see how. We've been using click-to-play safeguards on embedded content for years, and they have proved highly effective at stopping abusive or outright malicious content in Flash, Java applets, etc. But other software on the user's computer wasn't trying to work around those safeguards. That's the main attack vector, as I understand it. > That malicious sites are actively compromising user…

I still don't see how this is as complicated as these arguments suggest.

There is a known risk of Firefox being compromised by malicious addons, including those preinstalled by certain organisations. This risk is what is moderated by requiring addons to be signed and hard-coding a block. However, moderation is all this gains, because anyone who is preinstalling Firefox on a computer could still install a modified executable instead.

There is also a known risk of the user's security or privacy being compromised by visiting malicious websites that exploit weaknesses or vulnerabilities in Firefox. This risk is what is moderated by addons that block or otherwise interfere with undesirable content. It doesn't take any sort of hindsight to anticipate this; it is one of the major reasons people advocate blocker extensions, and this has been true for many years.

It is understandable that Mozilla would want to disrupt the former threat, but as I and others have explained, there are tried and tested ways they could do so that are no more vulnerable than the current approach yet would not suddenly remove all protection offered by addons against the latter threat without warning in the middle of a browsing session. The current heavy-handed approach is like building a secure home by making a concrete bunker with no doors and windows: the efforts to secure the addon system ultimately rendered the entire system useless.

Worse than that, though, the current strategy violates the basic principles that attract some users to Firefox in the first place, specifically its extensibility through addons and its relative respect for users' privacy and control of their own systems. The fact that Mozilla have so far shown little understanding of why some users would have a problem with this is regrettable, but perhaps they will come around with further thought after the event. However, the fact that there are people here still trying to defend the policy despite the highly visible train wreck that just happened seems very odd to me.

Re: Technical Details on the Recent Firefox Add-On Outage

#273
post #263

Earlier quoted context omitted.

> Sorry, but I don't really see how. We've been using click-to-play safeguards on embedded content for years, and they have proved highly effective at stopping abusive or outright malicious content in Flash, Java applets, etc. But other software on the user's computer wasn't trying to work around those safeguards. That's the main attack vector, as I understand it. > That malicious sites are actively compromising user…

I still don't see how this is as complicated as these arguments suggest. There is a known risk of Firefox being compromised by malicious addons, including those preinstalled by certain organisations. This risk is what is moderated by requiring addons to be signed and hard-coding a block. However, moderation is all this gains, because anyone who is preinstalling Firefox on a computer could still install a modified exe…

> However, moderation is all this gains, because anyone who is preinstalling Firefox on a computer could still install a modified executable instead.

Well, apparently that is a line that vendors are not prepared to cross.

> There is also a known risk of the user's security or privacy being compromised by visiting malicious websites that exploit weaknesses or vulnerabilities in Firefox.

When it comes to actual weaknesses or vulnerabilities, it seems clear to me that Mozilla should not rely on add-ons for patching those. But yes, blocker extensions still provide value; luckily, they are also still allowed.

> as I and others have explained, there are tried and tested ways they could do so that are no more vulnerable than the current approach yet would not suddenly remove all protection offered by addons against the latter threat without warning in the middle of a browsing session. The current heavy-handed approach is like building a secure home by making a concrete bunker with no doors and windows: the efforts to secure the addon system ultimately rendered the entire system useless.

You've said this before, so to prevent getting into a loop, I won't repeat my response :)

> Worse than that, though, the current strategy violates the basic principles that attract some users to Firefox in the first place, specifically its extensibility through addons and its relative respect for users' privacy and control of their own systems.

This I understand, and I wish it wasn't necessary too. I do think Mozilla has not shown little understanding - they've repeatedly explained how they are caught between a rock and hard place, and reached a different conclusion than you did, after weighing the pros and cons. That does not mean a lack of understanding of the cons, but merely that they did not outweigh the cons of the alternatives in their view.

This might simply be the result of different valuations of the pros and cons between you and Mozilla; given the amount of data and insight Mozilla has on the use of Firefox, I would also suggest to be open to the idea that there might be a lack of understanding on our side about the scale of the problem of malicious extensions.

Re: Technical Details on the Recent Firefox Add-On Outage

#275

Earlier quoted context omitted.

Imagine you have an old expired key... you take your new malicious extension, and sign it with the expired key and a time stamp that says it was signed at a time the key was still valid. Without some other verification mechanism, you can't tell the difference between this and an actual signature signed when the key WAS valid

Oh, the developers are signing their own extensions? I wasn't aware how it worked and I was missing that part, I thought Mozilla signed them on upload (and thus could trust itself to not be malicious).

Well, Mozilla does the signing, but we only know that to be true as long as they control the private key. The whole point of expirations is to make mitigate the risk of an older key being stolen (or cracked).

So yes, Mozilla signs the extensions, but that doesn't change the importance of keeping the private key private... that is HOW we know it is Mozilla doing the signing

Re: Technical Details on the Recent Firefox Add-On Outage

#276

Earlier quoted context omitted.

Hey sciurus, I just wanted to chime in way down deep in this comment chain because my thought only makes sense in the context of your comment right here. I think there may be a special mode of operation of Firefox that may need to be considered here. You said, "Even if you had no internet connection, your addons would have stopped working when the certificate expired." This seems like an unfortunate design flaw to me…

> I think it is a reasonable expectation that the marooned Firefox should continue to run indefinitely without failure. I personally agree this is a worthwhile goal. The blog post talks about "tracking the status of everything in Firefox that is a potential time bomb and making sure that we don’t find ourselves in a situation where one goes off unexpectedly." I expect once that is done we'll be positioned to evaluate…

Great! I think you understand me fully. Appreciate the willingness to explore such a strange but interesting freedom-related use case.

Re: Technical Details on the Recent Firefox Add-On Outage

#277

Earlier quoted context omitted.

The Studies mechanism is clear, public and fully transparent. You can see what they've done. You can see what they plan to do. It's all out in the open. At the end of the day the question is whether you can trust Mozilla. I trust them more than most entities, including many that push changes through apt-get.

Yes, Studies, that mechanism so open that end users didn't know about it until they found they were participating in some random Augmented Reality marketing collaboration with a TV show. I would love to see a link to the code related to various Studies. So far I've not been able to find any. All you seem to get told is the name of any studies you're in, and only if you go over to about:studies and go looking. This is…

You can see the bugs filed for past and upcoming studies at https://bugzilla.mozilla.org/buglist.cgi?list_id=14712186&re...

When the study involves an addon, I think the bug will link to its code.

(Disclosure: I work for Mozilla)

Re: Technical Details on the Recent Firefox Add-On Outage

#278

Earlier quoted context omitted.

Just to be clear: supported version or not, this sucks, and I hope we'll have a fix for you. I wanted to point out this wasn't remotely broken, however. Even if you had no internet connection, your addons would have stopped working when the certificate expired. (Disclosure: I work for Mozilla)

Well that's reassuring! :-) I've fixed it now anyway, in the way I specifically read that I wasn't meant to do. Maybe you guys could inform us unsupported old version users what we should do instead of waiting for an update that can't come? It's ok, thanks for your post. I know stuff is complicated and shit happens. I'm much more disappointed that the latest Firefox doesn't have a working session manager than I am ab…

In case it still helps you or someone else: A fix for Firefox 52 through 56 is available now at https://addons.mozilla.org/en-US/firefox/addon/disabled-add-...

(Disclosure: I work for Mozilla)

Re: Technical Details on the Recent Firefox Add-On Outage

#279

Earlier quoted context omitted.

Well that's reassuring! :-) I've fixed it now anyway, in the way I specifically read that I wasn't meant to do. Maybe you guys could inform us unsupported old version users what we should do instead of waiting for an update that can't come? It's ok, thanks for your post. I know stuff is complicated and shit happens. I'm much more disappointed that the latest Firefox doesn't have a working session manager than I am ab…

In case it still helps you or someone else: A fix for Firefox 52 through 56 is available now at https://addons.mozilla.org/en-US/firefox/addon/disabled-add-... (Disclosure: I work for Mozilla)

Great. Thank you.

Would you mind please telling whoever you need to that we need automatic session saving/restoring to work properly, and then lots of people like me will happily update.

Post reply on HN